U.S. patent application number 16/239993 was filed with the patent office on 2019-05-09 for spatially-segmented content delivery.
This patent application is currently assigned to Koninklijke KPN N.V.. The applicant listed for this patent is Koninklijke KPN N.V.. Invention is credited to Anton Havekes, Omar Aziz Niamut, Martin Prins, Ray van Brandenburg, Mattijs Oskar van Deventer.
Application Number | 20190141373 16/239993 |
Document ID | / |
Family ID | 46208606 |
Filed Date | 2019-05-09 |
![](/patent/app/20190141373/US20190141373A1-20190509-D00000.png)
![](/patent/app/20190141373/US20190141373A1-20190509-D00001.png)
![](/patent/app/20190141373/US20190141373A1-20190509-D00002.png)
![](/patent/app/20190141373/US20190141373A1-20190509-D00003.png)
![](/patent/app/20190141373/US20190141373A1-20190509-D00004.png)
![](/patent/app/20190141373/US20190141373A1-20190509-D00005.png)
![](/patent/app/20190141373/US20190141373A1-20190509-D00006.png)
![](/patent/app/20190141373/US20190141373A1-20190509-D00007.png)
![](/patent/app/20190141373/US20190141373A1-20190509-D00008.png)
![](/patent/app/20190141373/US20190141373A1-20190509-D00009.png)
![](/patent/app/20190141373/US20190141373A1-20190509-D00010.png)
View All Diagrams
United States Patent
Application |
20190141373 |
Kind Code |
A1 |
van Deventer; Mattijs Oskar ;
et al. |
May 9, 2019 |
Spatially-Segmented Content Delivery
Abstract
Methods and systems are described for processing spatially
segmented content originating from a content delivery network. One
method comprises a client in a media processing device receiving a
spatial manifest information comprising one or more spatial
representations of a source stream, each spatial representation
identifying one or more spatial segment streams, location
information for locating one or more delivery nodes in said content
delivery network and, optionally, position information for
stitching spatial segment frames in said segment streams into a
video frame for display; selecting one or more spatial segment
streams and on the basis of said spatial manifest information
requesting at least one delivery node in said content delivery
network to transmit said one or more selected spatial segment
streams to said client; and, receiving said one or more selected
spatial segment streams from said at least one delivery node.
Inventors: |
van Deventer; Mattijs Oskar;
(Leidschendam, NL) ; Niamut; Omar Aziz;
(Vlaardingen, NL) ; Havekes; Anton; (Schiedam,
NL) ; Prins; Martin; (The Hague, NL) ; van
Brandenburg; Ray; (The Hague, NL) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Koninklijke KPN N.V. |
Rotterdam |
|
NL |
|
|
Assignee: |
Koninklijke KPN N.V.
Rotterdam
NL
|
Family ID: |
46208606 |
Appl. No.: |
16/239993 |
Filed: |
January 4, 2019 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
15846891 |
Dec 19, 2017 |
|
|
|
16239993 |
|
|
|
|
14122198 |
Nov 25, 2013 |
9860572 |
|
|
PCT/EP2012/060801 |
Jun 7, 2012 |
|
|
|
15846891 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
H04L 29/06 20130101;
H04N 21/440263 20130101; H04N 21/2356 20130101; H04N 21/4728
20130101; H04N 21/6587 20130101; H04N 21/218 20130101; H04L 65/4084
20130101; H04N 21/234363 20130101; H04L 67/18 20130101 |
International
Class: |
H04N 21/2343 20060101
H04N021/2343; H04L 29/06 20060101 H04L029/06; H04N 21/6587 20060101
H04N021/6587; H04L 29/08 20060101 H04L029/08; H04N 21/235 20060101
H04N021/235; H04N 21/4728 20060101 H04N021/4728; H04N 21/218
20060101 H04N021/218; H04N 21/4402 20060101 H04N021/4402 |
Foreign Application Data
Date |
Code |
Application Number |
Jun 8, 2011 |
EP |
11169068.1 |
Claims
1. Method for processing spatially segmented content comprising: a
client receiving a spatial manifest data structure comprising one
or more spatial representations of a source stream, each spatial
representation identifying one or more spatial segment streams and
location information for locating one or more delivery nodes
associated with a content delivery network configured to transmit
said one or more spatial segment streams to said client selecting
one or more spatial segment streams associated with at least one of
said spatial representations and, on the basis of said location
information, requesting at least one delivery node associated with
said content delivery network to deliver said at least one selected
spatial segment stream to said client; and, receiving said one or
more selected spatial segment streams from said at least one
delivery node.
Description
FIELD OF THE INVENTION
[0001] The invention relates to spatially-segmented content
delivery and, in particular, though not exclusively, to methods and
systems for processing spatially-segmented content, a content
delivery network configured to distribute spatially-segmented
content, a client configured for processing spatially-segmented
content, a data structure for controlling spatially-segmented
content in a client or a content delivery network and a computer
program product using such method.
BACKGROUND OF THE INVENTION
[0002] New production equipment is starting to become available
that allows capturing and transmission of wide-angle, very high
resolution video content. The processing of such media files
requires high bandwidths (order of Gbps) and in order to exploit
the full information in such video stream large video screens,
typically the size of a video wall or a projection screen, should
be used.
[0003] When watching video on a relatively small display, a user is
not able not watch the entire video frame and all information
contained therein. Instead, based on his or her preference a user
will select a particular frame area, viewing angle or video
resolution by interacting with the graphical user interface of the
user equipment. Apart from the fact that current network capacity
cannot handle the large bandwidth demands of these beyond-HD
resolutions, it is simply not efficient to deliver content that
will not be displayed on the screen anyway.
[0004] Mavlankar et al. describe in the article "An interactive
region-of-interest video streaming system for online lecturing
viewing", proceeding of ICIP 2010, p. 4437-4440 a so-called tiled
video system wherein multiple resolution layers of a video content
file are generated using a server-based interactive
Region-of-Interest encoder. Every layer is divided into multiple
spatial segments or tiles, each of which can be encoded and
distributed (i.e. streamed) independently from each other.
[0005] A video client may request a specific spatial region, a
Region of Interest (ROI), and a server may subsequently map the ROI
request to one or more segments and transmit a selected group of
tiled streams to the client, which is configured to combine the
tiled streams into one video. This way bandwidth efficient
user-interactions with the displayed content, e.g. zooming and
panning, may be achieved without requiring compromises with respect
to the resolution of the displayed content. In such scheme however
fast response and processing of the tiled streams is necessary in
order to deliver quality of services and optimal user
experience.
[0006] When implementing such services in existing large-scale
content delivery schemes problems may occur. Typically, a content
provider does not directly deliver content to a consumer. Instead,
content is sent to a content distributor, which uses a content
delivery network (CDN) to deliver content to the consumer. CDNs are
not arranged to handle spatially segmented content. Moreover, known
solutions for delivering spatially segmented content to a client
(such as described in the article of Mavlankar) are not suitable
for use with existing CDNs. Implementing a known server-based
solution for resolving ROI requests in a CDN in which spatially
segmented content may be stored at different locations, does not
provide a scalable solution and may easily lead undesired effects
such as overload and delay situations which may seriously
compromise the realisation of such service.
[0007] Furthermore, the server-based ROI to tile mapping approach
described by Mavlankar does not enable the possibility of a client
accessing content through two separate access methods/protocols,
such as a scenario in which some spatial segments (segment streams)
may be received using broadcast or multicast as access technology
(medium) and some other spatial segments may be received using for
instance unicast as access technology. The server-based ROI mapping
mechanism of Mavlankar will have no knowledge of the exact
client-side capabilities for receiving segmented content via
different access technologies and or access protocols.
[0008] Moreover, in order to handle and display a video stream on
the basis of several spatially-segmented streams, specific
functionality needs to be implemented in the client in order to
enable the display of seamless video images on the basis of a
number of spatial segment streams and to enable user-interaction
with the displayed content.
[0009] Hence, there is a need in the prior art for efficient
spatially-segmented content delivery to clients. In particular,
there is a need for methods and systems for spatially-segmented
content delivery systems which allow content delivery without
requiring changes to conventional CDN infrastructure, and which may
also provide a scalable and flexible solution for users interacting
with high-resolution video content.
SUMMARY OF THE INVENTION
[0010] It is an object of the invention to reduce or eliminate at
least one of the drawbacks known in the prior art and to provide in
a first aspect of the invention a method for for processing
spatially segmented content comprising: a client receiving a
spatial manifest data structure comprising one or more spatial
representations of a source stream, each spatial representation
identifying one or more spatial segment streams and location
information for locating one or more delivery nodes, such one or
more delivery nodes preferably associated with a content delivery
network configured to transmit said one or more spatial segment
streams to said client; selecting one or more spatial segment
streams associated with at least one of said spatial
representations and, on the basis of said location information,
requesting at least one delivery node, preferably associated with
said content delivery network, to deliver said at least one
selected spatial segment stream to said client; and,
receiving said one or more selected spatial segment streams from
said at least one delivery node.
[0011] The spatial manifest data structure (file) allows efficient
client-based video manipulation actions, including zooming, tilting
and panning of video data wherein the functionality in the network,
in particular in the CDN, can be kept simple when compared to prior
art solutions.
[0012] In one embodiment said spatial manifest data structure may
further comprise positioning information for stitching spatial
segment frames in said spatial segment streams into a video frame
for display.
[0013] In another embodiment said spatial manifest data structure
may comprises at least a first spatial representation associated
with a first resolution version of said source stream and at least
a second spatial representation associated with a second resolution
version of said source stream.
[0014] In an embodiment said spatial manifest data structure may
comprise a first spatial representation associated with a first
number of segment streams and said second spatial representation
comprises a second number of segment streams.
[0015] In another embodiment said method may further comprise:
buffering said one or more selected spatial segment streams;
receiving synchronization information associated with said one or
more spatial segment streams; synchronizing said one or more
requested spatial segment streams on the basis of said
synchronization information.
[0016] In one embodiment said method may comprise: on the basis of
location information stitching said spatially aligned spatial
segment frames into a continuous video frame for play-out.
[0017] In a further embodiment said one or more requested spatial
segment streams may be transmitted to said client using an adaptive
streaming protocol, preferably an HTTP adaptive streaming protocol,
wherein at least part of said spatial segment streams is divided in
time segments.
[0018] In yet further embodiment said one or more segment streams
may be transmitted to said client using an RTP streaming protocol
and wherein synchronization information is transmitted to said
client in an RTSP signaling message.
[0019] In a further embodiment said spatial manifest data structure
contains references to one or more access technologies or protocols
(e.g. unicast, multicast, broadcast), said references preferably
associated with one or more, preferably spatial, segment
streams.
[0020] In a further embodiment said one or more segment streams may
be transmitted to the client, using one or more of said access
technologies and/or access protocols. For example, some segments
streams may be transmitted using a broadcast medium (e.g.
broadcast, multicast) as access technology, whereas some other
segment streams may be transmitted using unicast as access
technology.
[0021] In yet another embodiment the spatial manifest data
structure (file) may be received, preferably by the clientover a
broadcast medium, such as multicast or broadcast.
[0022] In a further embodiment the spatial manifest data structure
(file) may be transmitted to multiple receivers (e.g. clients)
simultaneously over a broadcast medium (e.g. simulcast).
[0023] In a further aspect the invention may relate to a method for
user-interaction with spatially segmented content displayed to a
user by a client in media processing device, said method
comprising: receiving a spatial manifest data structure comprising
one or more spatial representations of a source stream, each
spatial representation identifying one or more spatial segment
streams and location information for locating one or more delivery
nodes associated with a content delivery network configured to
transmit said one or more spatial segment streams to said client;
on the basis of said location information said spatial manifest
data structure, retrieving one or more spatial segment streams
associated with a first spatial representation from said content
delivery network and displaying said one or more spatial segment
streams; receiving user input associated with a region of interest,
said region of interest defining a selected area of said displayed
spatially-segmented content; on the basis of positioning
information in said spatial segment data structure and said region
of interest determining a second spatial representation in
accordance with one or more media processing criteria, preferably
said processing criteria including at least one of bandwidth,
screen resolution and/or encoding quality.
[0024] In one embodiment said method may further comprise for said
one or more spatial representations determining the spatial segment
frames overlapping said selected region of interest
[0025] In a further aspect the invention may relate to a method for
ingesting spatially segmented content originating from at least one
source node into a first content delivery network, preferably said
at least one source node being associated with a second content
delivery network, the method comprising: receiving a spatial
manifest data structure comprising one or more spatial
representations of a source stream, each spatial representation
identifying one or more spatial segment streams and location
information for locating one or more delivery nodes associated with
a content delivery network configured to transmit said one or more
spatial segment streams to said client;, said at least one source
node configured to transmit said one or more spatial segment
streams to said first content delivery network, distributing said
at least one spatial segment stream to one or more delivery nodes
in said first content delivery network; and, updating said spatial
manifest information by adding location information associated with
said one or more delivery nodes to said spatial manifest
information.
[0026] In another aspect the invention may relate to a content
distribution network arranged for distribute spatially segmented
content, said network comprising: a control function, an ingestion
node and one or more delivery nodes; wherein said control function
is configured to receive a spatial manifest data structure
comprising one or more spatial representations of a source stream,
each spatial representation identifying one or more spatial segment
streams and location information for locating one or more delivery
nodes associated with a content delivery network configured to
transmit said one or more spatial segment streams to said client;
and wherein said control function is configured to request said one
or more spatial segment streams from said one or more source nodes
on the basis of said location information; wherein said ingestion
node is configured to receive one or more spatial segment streams
from said one or more source nodes; and, wherein said one or more
delivery nodes are configured to receive said ingested one or more
spatial segment streams.
[0027] In a further aspect, the invention may relate to a control
function configured to receive a spatial manifest data structure
comprising one or more spatial representations of a source stream,
each spatial representation identifying one or more spatial segment
streams and location information for locating one or more delivery
nodes associated with a content delivery network configured to
transmit said one or more spatial segment streams to said client;
and configured to request said one or more spatial segment streams
from said one or more source nodes on the basis of said location
information.
[0028] The invention may further relate to a client for processing
spatially segmented content, wherein the client is configured for:
receiving a spatial manifest data structure comprising one or more
spatial representations of a source stream, each spatial
representation identifying one or more spatial segment streams and
location information for locating one or more delivery nodes
associated with a content delivery network configured to transmit
said one or more spatial segment streams to said client; selecting
one or more spatial segment streams and on the basis of said
location information requesting at least one delivery node in said
content delivery network to transmit said one or more selected
spatial segment streams to said client; and, receiving said one or
more selected spatial segment streams from said at least one
delivery node.
[0029] In one embodiment said client is further configured for
buffering said one or more selected spatial segment streams;
receiving synchronization information associated with said one or
more spatial segment streams; and synchronizing said one or more
requested spatial segment streams on the basis of said
synchronization information.
[0030] In another embodiment said client is further configured for
stitching spatial segment frames on position information in said
spatial manifest data structure.
[0031] The invention may further relate to a spatial manifest data
structure for use in a client of a play-out device, preferably a
client as described above, said data structure controlling the
retrieval and display of spatially segmented content originating
from a content delivery network, wherein said data structure
comprises one or more spatial representations of a source stream,
each spatial representation identifying one or more spatial segment
streams, location information for locating one or more delivery
nodes in said content delivery network and, optionally, position
information for stitching spatial segment frames in said segment
streams into a video frame for display.
[0032] The invention may also relate to a spatial manifest data
structure for use by a content delivery network, preferably a
content delivery network as described above, said spatial manifest
data structure controlling the ingestion of spatially segmented
content, wherein said data structure comprises one or more spatial
representations of a source stream, each spatial representation
identifying one or more spatial segment streams, location
information for locating said one or more source nodes and,
optionally, position information for stitching spatial segment
frames in said segment streams into a video frame for display.
[0033] The invention also relates to a computer program product
comprising software code portions configured for, when run in the
memory of computer, executing at least one of the method steps as
described above.
[0034] The invention will be further illustrated with reference to
the attached drawings, which schematically will show embodiments
according to the invention. It will be understood that the
invention is not in any way restricted to these specific
embodiments.
BRIEF DESCRIPTION OF THE DRAWINGS
[0035] FIG. 1 depicts a segmented content delivery system according
to an embodiment of the invention.
[0036] FIG. 2A-2E depict the generation and processing of spatially
segmented content according to various embodiments of the
invention.
[0037] FIG. 3 depicts a SMF data structure according to one
embodiment of the invention.
[0038] FIG. 4 depicts an example of part of an SMF according to an
embodiment of the invention.
[0039] FIG. 5A-5C depict the process of a user interacting with a
content processing device which is configured to process and
display spatially segmented content according to an embodiment of
the invention.
[0040] FIGS. 6A and 6B depict flow diagrams of processes for
delivering spatially segmented content to a client according to
various embodiments of the invention.
[0041] FIG. 7A-7C depict parts of SMFs according to various
embodiments of the invention.
[0042] FIGS. 8A and 8B illustrate the use of different coordinate
systems for defining the position of spatial segment frames in an
image frame.
[0043] FIG. 9 depicts a schematic of a media engine according to an
embodiment of the invention.
[0044] FIGS. 10A and 10B illustrate synchronized and stitched
spatial segment frames based on different streaming protocols.
[0045] FIG. 11 depicts a process for determining synchronization
information for synchronizing spatially segment content according
to an embodiment of the invention.
[0046] FIG. 12 depicts an example of a client requesting content
from two different delivery nodes.
[0047] FIG. 13 depicts a process for improving the user-interaction
with the spatially segmented content according, to an embodiment of
the invention.
DETAILED DESCRIPTION
[0048] FIG. 1 depicts a content delivery system 100 for delivering
spatially segmented content to a client according to one embodiment
of the invention. The content delivery system may comprise a
content delivery network (CDN) 102, a content provider system (CPS)
124 and a number of clients 104.sub.1-N. The CPS may be configured
to offer content, e.g. a video title, to customers, which may
purchase and receive the content using a client hosted on a
terminal.
[0049] A terminal may generally relate to a content processing
device, e.g. a (mobile) content play-out device such as an
electronic tablet, a smart-phone, a notebook, a media player, etc.
In some embodiment, a terminal may be a set-top box or content
storage device configured for processing and temporarily storing
content for future consumption by a content play-out device.
[0050] Delivery of the content is realized by the delivery network
(CDN) comprising delivery nodes 108.sub.1-M and at least one
ingestion node 108.sub.0. A CDN control function (CDNCF) 111 is
configured to receive content from an external content source, e.g.
a content provider 110 or another CDN (not shown). The CDNCF may
further control the ingestion node to distribute one or more copies
of the content to the delivery nodes such that throughout the
network sufficient bandwidth for content delivery to a terminal is
guaranteed. In one embodiment, the CDN may relate to a CDN as
described in ETSI TS 182 019.
[0051] A customer may purchase content, e.g. video titles, from the
content provider by sending a request 118 to a web portal (WP) 112,
which is configured to provide title references identifying
purchasable content. In response to the request, a client may
receive at least part of the title references from the WP and
location information, e.g. an URL, of a CDNCF of a CDN, which is
able to deliver the selected content. The CDNCF may send a client
location information associated with at least one delivery node in
the CDN, which are able to deliver the selected content to the
client. Typically, the CDNCF will select a delivery node in the
CDN, which is best suited for delivering the selected content to
the client. Criteria for selecting a delivery node may include
location of the client and the processing load of the delivery
nodes.
[0052] A client may contact the delivery node in the CDN using
various known techniques including a HTTP and/or a DNS system.
Further various streaming protocols may be used to deliver the
content to the client. Such protocols may include HTTP and RTP type
streaming protocols. In a preferred embodiment an adaptive
streaming protocol such as HTTP adaptive streaming (HAS), DVB
adaptive streaming, DTG adaptive streaming, MPEG DASH, ATIS
adaptive streaming, IETF HTTP Live streaming and related protocols
may be used.
[0053] The CDN is configured to ingest so-called spatially
segmented content and distribute the spatially segmented content to
a client. Spatially segmented content may be generated by spatially
dividing each image frame of a source stream, e.g. an MPEG-type
stream, into a number spatial segment frames. A time sequence of
spatial segment frames associated with one particular area of an
image frame may be organized in a separate spatial segment stream.
The spatial segment frames in a spatial segment stream may be
formatted according to a known transport container format and
stored in a memory as a spatial segment file.
[0054] In order to handle ingestion of spatially segmented content
and to delivery spatially segmented content to client the CDNCF
uses a special data structure, which hereafter will be referred to
as a spatial manifest file (SMF). A SMF may be stored and
identified using the zzzz.smf file extension. The use of the SMF in
CDN-based spatially segmented content delivery to a client will be
described hereunder in more detail.
[0055] FIG. 2A-2E depict the generation and processing of spatially
segmented content according to various embodiments of the
invention.
[0056] FIG. 2A schematically depicts an exemplary process for
generating spatially segmenting content. A content provider may use
the process in order to delivery such content via a CDN to a
client. The process may be started by an image processing function
generating one or more lower resolution versions of the
high-resolution source stream 220. Thereafter, each image frame 222
in a lower resolution video stream is divided in separate, spatial
segments frames 224.sub.1-16, wherein each spatial segment frame is
associated with a particular area in the image frame. All spatial
segment frames required for constructing an image frame of the
original source stream is hereafter referred to as a spatial
segment group.
[0057] In one embodiment, a spatial segment frame may comprise a
segment frame header comprising spatial segment information, e.g.
information regarding the particular position in an image frame it
belongs to, sequence information (e.g. a time stamp) and resolution
information.
[0058] A sequence of spatial segment frames associated with one
particular area of an image frame of the source stream are
subsequently encoded in a particular format, e.g. AVI or MPEG, and
ordered in a spatial segment stream 226.sub.1-16 format. Each
spatial segment stream may be stored as a spatial segment file,
e.g. xxxx.m2ts or xxxx.avi. During segmentation, the image
processing function may generate information for establishing the
spatial relation between spatial segment streams and location
information indicating where the spatial segment streams may be
retrieved from.
[0059] It is submitted that the spatial division of image frames
into a segment group is not limited to those schematically depicted
in FIG. 2A. An image frame may be spatially divided into a segment
group defining a matrix of spatial segment frames of equal
dimension. Alternatively, in other embodiment, spatial segment
frames in a segment group may have different segment frame
dimensions, e.g. small segment frames associated with the center of
the original image frame and larger sized spatial segment frames at
the edges of an image frame.
[0060] The spatial segmentation process depicted in FIG. 2A may be
repeated for different resolution versions of the source stream.
This way, on the basis of one source stream, multiple spatial
representations may be generated, wherein each spatial
representation may be associated with a predetermined spatially
segmented format of the source file.
[0061] The information establishing the spatial and/or temporal
relation between spatial segment streams in each spatial
representation and the location information, e.g. URLs, for
retrieving the spatial segment streams may be hierarchically
ordered and stored as a spatial manifest file (SMF). The structure
of the SMF will be discussed in detail with reference to FIG.
3.
[0062] FIG. 2B depicts at least part of a CDN 200 according to one
embodiment of the invention in more detail. The CDN may comprise
one or more delivery nodes 202.sub.1-2, each comprising a
controller 204.sub.1-2 and a delivery storage 206.sub.1-2. An
ingestion node 208 is configured to receive spatially segmented
content from a content processing entity 210, e.g. a content
provider or another CDN, under control of the CDNCF.
[0063] In order to ingest spatially segmented content, a CDNCF 212
may receive a SMF from the content processing entity wherein the
SMF comprises location information, e.g. URLs, where the spatially
segmented content associated with the content processing entity is
stored. Based on the location information in the SMF, the CDNCF may
then initiate transport of spatial segment streams to the ingestion
node, which may temporarily store or buffer the ingested content in
an ingestion cache 214.
[0064] The CDNCF may further comprise a content deployment function
216 for distributing copies of spatial content segment streams to
one or more delivery nodes in accordance with a predetermined
distribution function. The function may use an algorithm, which
allows distribution copies of the spatially segmented content over
one or more delivery nodes of the CDN based on the processing load,
data traffic and/or geographical proximity information associated
with the delivery nodes in the CDN.
[0065] Hence, when receiving a content request from a client, the
CDNCF may select one or more delivery nodes, which are e.g. close
to the client and do not experience a large processing load. The
different spatial segment streams associated with one SMF may be
distributed to different delivery nodes in order to improve the
efficiency of the CDN. For example, the more-frequently requested
spatial segment streams positioned in the center of an image frame
may be stored one multiple delivery nodes, while the
less-frequently requested spatially segment streams positioned at
the edge of an image frame may be stored on one delivery node.
[0066] Once the CDNCF has distributed ingested spatially segmented
content to the delivery nodes, the CDNCF may modify location
information in the SMF such that the modified SMF comprises
references (URLs) to the delivery nodes to which the spatial
segment streams were distributed. The thus modified SMF is stored
in a SMF storage 218 with the CDNCF.
[0067] FIGS. 2C and 2D depict a flow chart and a sequence diagram
respectively associated with a process of ingesting spatially
segmented content in accordance with one embodiment of the
invention. As shown in FIG. 2C, the process may start with the
CDNCF receiving a request for content ingestion originating from a
source, e.g. a content provider, a CDN or another content
processing entity (step 230). The request may comprise location
information, e.g. an URL, and a content identifier. In response,
the CDNCF receives a file (step 232), which may either relate to a
SMF (zzzz.smf) or a conventional, non-segmented content file.
[0068] If the CDNCF receives an SMF, it may execute ingestion of
spatially segmented content on the basis of the received SMF (step
234). Execution of the ingestion process, may include parsing of
the SMF (step 236) in order to retrieve the location information,
e.g. URLs, associated with the spatial segment streams and
instructing the ingestion node to retrieve the respective spatial
segment streams and temporarily storing the ingested streams in the
ingestion cache (step 238).
[0069] On the basis of processing load, data traffic and/or
geographical proximity information associated with the delivery
nodes in the CDN, the CDN deployment function may then distribute
the spatial segment files over one or more predetermined content
delivery nodes (step 240) and subsequently update the SMF by adding
location information associated with the new locations of different
delivery nodes where the spatial different segment files are stored
(step 242). Optionally, URL may also send to the source (step
254).
[0070] The CDNCF may store the updated (modified) SMF with at least
one of the delivery nodes (step 244) and generate a new entry in a
SMF table associated with the CDNCF in order to store the location
where the updated SMF may be retrieved for future use (step 246 and
248). Optionally, URL for locating the SMF may also be sent to the
source (step 254).
[0071] If the source ingestion request of the source is associated
with conventional non-segmented content, the content may be
temporarily stored in the ingestion storage, where after the CDN
deployment function distributes the content to on or more delivery
nodes. (step 250). A URL associate with the stored file may be
generated and stored with the CDNCF for later retrieval (step 252).
Optionally, URL for locating the content may also be sent to the
source (step 254).
[0072] The sequence diagram of FIG. 2D depicts on embodiment of a
process for ingesting spatially segmented content in more detail.
In this example, the CDNCF and the source may use an HTTP-based
protocol for controlling the content ingestion. The process may
start with the source sending a content ingestion request to the
CDNCF wherein the request comprises an URL pointing to a SMF named
Movie-4.smf (step 260). The CDNCF may retrieve the SMF (steps 262
and 264) and parse the SMF (step 268) in order to retrieve the
spatial segment file locations (e.g. in the form of one or more
URLs).
[0073] Using the URLs, the CDNCF instructs the CDN ingestion node
to fetch the spatial segment streams identified in the SMF, e.g. a
source storage cache (step 270). After having received the spatial
segment streams (steps 272-278), the ingestion node may notify the
CDNCF that the content has been successfully retrieved (step 280).
The CDNCF then may distribute copies of the ingested streams to the
delivery nodes in the CDN. For example, on the basis of processing
load, data traffic and/or geographical proximity information, the
CDN deployment function may decide to distribute copies of the four
spatially segment streams Movie-4-1.seg, Movie-4-2.seg,
Movie-4-3.seg and Movie-4-4.seg to distribution node ND2 and the
spatially segment streams Movie-4-1.seg and Movie-4-3.seg to DN1
(step 281).
[0074] After distribution to the delivery nodes, the CDNCF may
update the SMF by inserting the location of the delivery nodes (in
this case the URLs of DN1 and D2) from which the spatially segment
streams may be retrieved into the SMF (step 282). The updated SMF
may be stored with the spatially segment streams (step 284 and
288). The location of the SMF may be stored in a SMF table
associated with the CDNCF (step 286). Optionally, the CDNCF may
send a response comprising location information associated with the
updated SMF to the source indicating successful content ingestion
by the CDN (step 290).
[0075] FIG. 2E depicts an example of part of an SMF before and
after ingestion of spatially segmented content as described with
reference to FIGS. 2C and 2D. The SMF 292 received from the source
may comprise a content identifier, information regarding the
segmentation, e.g. type of segmentation, the number of segments and
the spatial relation between spatial segments from the different
spatial segment files. For example the content of the exemplary SMF
in FIG. 2E indicates that the spatially segment content was
generated on the basis of a source file Movie_4. The SMF further
indicates that four separate spatial segment streams Movie-4-1.seg,
Movie-4-2.seg, Movie-4-3.seg and Movie-4-4.seg are stored at a
certain location cache.source.com/res/ in the source domain.
[0076] After ingestion of the spatial segment files, the CDNCF may
generate an updated SMF 294 comprising updated spatial segment
locations in the form of URLs pointing towards a first delivery
node cache_1.cdn_A/res/ comprising spatial segment streams
Movie-4-1.seg and Movie-4-3.seg and a second delivery node
cache_2.cdn_A/res/ comprising all four spatial segment streams
Movie-4-1.seg, Movie-4-2.seg, Movie-4-3.seg and Movie-4-4.seg.
[0077] Hence, from the above it follows that the SMF allows
controlled ingestion of spatially segmented content into a CDN.
After ingestion, the SMF identifies the locations in the CDN where
spatial segment streams may be retrieved. As will be described
hereunder in more detail, the SMF may further allow a client to
retrieve spatially segmented content and interact with it. In
particular, the SMF allows client-controlled user-interaction with
displayed content (e.g. zooming, panning and tilting
operations).
[0078] FIG. 3 schematically depicts a SMF data structure 300
according to one embodiment of the invention. An SMF may comprise
several hierarchical data levels 302,308,318,328, wherein a first
level 302 may relate to spatial composition information defining
one or more so-called spatial representations 306.sub.1-3 of a
source stream (e.g. source1.m2ts). Typically, the source stream may
be a high-resolution and, often, wide-angle video stream.
[0079] The next data level 308 may relate to spatial representation
information defining a spatial representation. In particular,
spatial representation information may define the spatial relation
between a set of spatial segment streams associated with a
particular version, e.g. a resolution version, of the source
stream.
[0080] The spatial representation information may comprise a set of
spatial segment stream instances 312.sub.1-4 arranged in a spatial
map 311. In one embodiment, the map may define a matrix (comprising
rows and columns) of spatial segment stream instances, which allow
construction of images comprising at least a part of the content
carried by the source stream.
[0081] A spatial segment stream instance may be associated with a
segment position 313.sub.1-4 defining the position of a segment
within a positional reference system. For example, FIG. 3 depicts a
matrix of spatial segment stream instances, wherein the spatial
segment stream instances associated with segment positions (1,1)
and (1,2) may refer to content associated with the top left and top
right areas of video images in the source stream. On the basis of
these two segment stream instances, video images may be constructed
comprising content associated with the top half of the video images
in the source file.
[0082] The spatial representation information may further comprise
a source resolution 310 indicating the resolution version of the
source stream, which is used to generate the spatial segment
streams referred to in the spatial map. For example, in FIG. 3 the
spatial segment streams referred to in the spatial map may be
generated on the basis of a 4096.times.2160 resolution version of
the source stream.
[0083] The spatial representation information may further comprise
a segment resolution 314 (in the example of FIG. 3 a
2048.times.1080 resolution) defining the resolution of the spatial
segment streams referred to in the spatial map and a segment
bit-stream rate 316 defining the bitrate at which spatial segment
streams in the spatial map are transmitted to a client.
[0084] A next data level 318 may relate to spatial segment
information, i.e. information associated with a spatial segment
stream instance in the spatial map. Spatial segment information may
include position coordinates associated with spatial segment frames
in a spatial segment stream 324. The position coordinates may be
based on an absolute or a relative coordinate system. The position
coordinates of the spatial segment frames in the spatial segment
streams referred to in the spatial map may be used by the client to
spatially align the borders of neighboring spatial segment frames
into a seamless video image for display. This process is often
referred to as "stitching".
[0085] Spatial segment information may further include one or more
spatial segment stream locators 326,328 (e.g. one or more URLs) for
locating delivery nodes, which are configured to transmit the
particular spatial segment stream to the client. The spatial
segment information may also comprise protocol information,
indicating which protocol (e.g. RTSP or HTTP) is used for
controlling the transmission of the streams to the client, and
synchronization information 330,332 for a synchronization module in
the client to synchronize different segment streams received by a
client.
[0086] The above-mentioned stitching and synchronization process
will be described hereunder in more detail with reference to FIG.
9-11.
[0087] The SMF may further comprise inter-representation
information 334 for relating a spatial segment stream in one
spatial representation to spatial segment stream in another spatial
representation. The inter-representation information allows a
client to efficiently browse through different representations.
[0088] FIG. 4 depicts an example of part of an SMF defining a set
of spatially segmented video representations of a (high resolution)
source file using (pseudo-)XML according to one embodiment of the
invention. This SMF defines different spatial representations (in
this case three) on the basis of the source file source1.m2ts.
[0089] The first spatial representation in the SMF of FIG. 4
defines a single, 1024.times.540 pixel resolution non-segmented
video stream (segment_mode==OFF). A client may retrieve the stream
using the URL rtsp://video.content.com/full_stream.avi defined in
the first spatial representation.
[0090] A second spatial representation in the SMF is associated
with a 4096.times.2160 resolution source stream, which is spatially
segmented in a 2.times.2 matrix (NumberOfColumns, NumberOfRows)
comprising four spatial segment stream instances, each begin
associated with a separate spatial segment stream of a
2048.times.1080 segment resolution and each being accessible by a
URL. For example, the first segment stream associated with segment
location (1,1) may be requested from the RTSP server using the URL
rtsp://video.content.com/stream1_1.avi.
[0091] The SMF may define synchronization information associated
with a spatial segment stream. For example, the SMF in FIG. 4
defines a predetermined RTP offset value for the RTSP server, which
is configured to transmit the spatial segment stream stream1_1.avi
to the client. The RTSP server uses the RTP protocol for streaming
the spatial segment stream to the client. Normally, consecutive RTP
packets in an RTP stream are time-stamped by the RTSP server using
a random offset value as a starting value. Knowledge about the
initial offset value is needed in order to synchronize different
spatial segment streams at the client.
[0092] Hence, in order to enable the client to synchronize
different RTP streams originating from different RTSP servers, the
client may send a predetermined RTP offset value in the requests to
the different RTSP servers, instructing each server to use the
predetermined RTP offset in the request as the starting value of
the RTP time stamping. This way, the client is able to determine
the temporal relation between the time-stamped RTP streams
originating from different RTSP servers.
[0093] A further, third spatial representation in the SMP may
relate to a 7680.times.4320 resolution version of the source
stream, which is segmented into a matrix comprising 16 spatial
segment stream instances, each being associated with a
1920.times.1080 resolution spatial segment stream and each being
accessible via predetermined URL.
[0094] Preferably, the resolution and aspect ratio of the spatially
segment streams is consistent with common video encoding and
decoding parameters, such as macro block and slice sizes. In
addition, the resolutions of the spatial segment streams should be
large enough as to retain a sufficient video coding gain.
[0095] FIG. 5A-5C depict the process of a user interacting with a
content processing device which is configured to process and
display spatially segmented content according to an embodiment of
the invention.
[0096] FIG. 5A depicts a content processing device 501 comprising a
client 502 configured to directly retrieve spatial segment streams
from one or more delivery nodes 504.sub.1,504.sub.2. Each delivery
node may comprise a controller 522.sub.1,522.sub.2 and a delivery
storage 524.sub.1,524.sub.2 for storing spatial segment
streams.
[0097] The display of the content processing device may be
configured as an interactive graphical user interface (GUI) 509
allowing both display of content and reception of user input 520.
Such user input may relate to user-interaction with the displayed
content including content manipulation actions such as zooming,
tilting and panning of video data. Such GUIs include displays with
touch-screen functionality well known in the art.
[0098] The client may comprise a control engine 505 configured for
receiving a SMF 503 from a CDN and for controlling a media engine
507 and an access engine 508 on the basis of the information in the
SMF and user input received via the GUI. Such user input may relate
to a user-interaction with displayed content, e.g. a zooming
action. Such user-interaction may initiate the control engine to
instruct the access engine (e.g. an HTTP or RTSP client) to
retrieve spatially segmented content using location information
(URLs) in the SMF.
[0099] FIG. 5B schematically depicts a user interacting with the
client through the GUI in more detail. In particular, FIG. 5A
schematically depicts a user selecting a particular area 512 (the
region of interest or ROI) of the display 514 as schematically
depicted in FIG. 5B. On the basis of the size and coordinates of
the selected ROI, the control engine may select an appropriate
spatial representation from the SMF. A process for selecting an
appropriate spatial representation will be described in more detail
with reference to FIG. 5C.
[0100] In the example of FIG. 5B the control engine may have
selected a spatial representation from the SMF comprising a matrix
of 16 spatial segment stream instances wherein each spatial segment
stream instance relates to content associated with a particular
area in the display (as schematically illustrated by the dotted
lines in display 514). The control engine may then determine the
overlap between the ROI and the spatial areas associated with the
spatial segment stream instances of the selected spatial
representation.
[0101] This way, in the example of FIG. 5B, the control engine may
determine that four spatial segment stream instances (gray-shaded
spatial segment stream instances S.sub.2,1; S.sub.3,1; S.sub.2,2;
S.sub.3,2) overlap with the ROI. This way four spatial segment
streams 515.sub.1-4 are identified by the control engine that allow
construction of a new video stream 516 displaying the content in
the ROI 512 to the user. The identified spatial segment stream
instances comprising the ROI are hereafter referred to as a spatial
segment subgroup.
[0102] With reference to FIG. 5A, after determining the spatial
segment subgroup, the control engine 505 may instruct the access
engine to request access to the delivery nodes configured to
deliver the identified spatial segment streams (in this example two
delivery nodes were identified). The access engine may then send a
request 511.sub.1-2 to the controller of delivery nodes 504.sub.1
and 504.sub.2 to transmit the spatial segment streams to the client
(in this example each delivery node delivers two spatial segment
streams 513.sub.1-2and 515.sub.1-2to the client). The access agent
508 may forward the transmitted spatial segment streams to the
media engine 507, which may comprise processing functions for
buffering, synchronizing and decoding the spatial segment streams
of the spatial segment subgroup and stitching decoded spatial
segment frames to video images comprising the ROI.
[0103] FIG. 5C depicts a flow diagram for user-interaction and
play-out of spatially segmented content on the basis of a SMF
according to one embodiment of the invention. In particular, FIG.
5C depicts a flow diagram wherein a user initiates a zooming action
by selecting a ROI.
[0104] The process may start with a client, in particular the
control engine in the client, receiving a SMF and selecting one of
the spatial representations as a default, e.g. a low bandwidth
unsegmented version of the source stream (step 530). Further, the
control engine may set the ROI to a default value, typically a
value corresponding to full display size (step 532). If the control
engine receives a ROI request, it may put all or at least part of
the spatial representations on a temporary list and mark the
spatial representations as "unchecked" (step 534). The control
engine may then start a process wherein an unchecked spatial
representation is taken from the list (step 536) and marked as
checked (step 538). On the basis of the SMF, the control engine may
then determine whether for the selected spatial representation a
spatial segment subgroup exists, i.e. a spatial segment stream
instances associated comprising the ROI (step 540)
[0105] Using the spatial representation information and the
associated spatial segment information, the control engine may
further verify whether an identified spatial segment subgroup
matches the requirements for play-out, e.g. bandwidth, codec
support, etc. (step 542). If the play-out requirements are met, the
next "unchecked" spatial representation is processed. If the
play-out requirements are not met, the control engine may remove
the spatial representation from the list (step 546). Thereafter the
next unchecked spatial representation is processed until all
spatial representations on the list are processed (step 544).
[0106] Then, after having checked all spatial representations on
the list, the client may check the list is empty (step 547). If the
list is empty, the client may determine whether a previous ROI and
spatial representation is available for selection (step 548). If
so, that previous ROI may be selected for play-out (step 549). If
no previous ROI is available, an error message may be produced
(step 550). Alternatively, if the list with checked spatial
representations is not empty, the content agent may select a
spatial representation from the list that best matches the play-out
criteria, e.g. screen resolution, quality, encoding, bandwidth,
etc., and subsequently changes to that selected spatial
representation (step 552).
[0107] The control engine may then execute a process for play-out
of the spatially segmented streams using the newly selected spatial
representation (step 554). In that case, the control engine may
determine the spatial segment subgroup (step 556) and the locations
information where the spatial segment streams in the spatial
segment subgroup may be retrieved (step 558). After retrieving the
spatial segment streams (step 560) they are processed for play-out
(step 562), i.e. buffered, synchronized and stitched in order to
form a stream of video images comprising the ROI.
[0108] During play-out of the spatially segmented content, the
control engine may be configured to detect a further ROI request,
e.g. a user selecting a further ROI via the GUI. If such signal is
detected (step 564), the above-described process for building and
checking a list of suitable spatial representations may be started
again.
[0109] The spatial representations in the SMF thus allow a user to
select different ROIs and to zoom in a video image by having the
client to select different spatial segment streams defined in the
one or more spatial representations in the SMF. The SMF allows a
client to directly request spatially segmented content from
delivery nodes in the CDN. Such direct retrieval provides fast
access to the content, which is required necessary in order to
deliver quality of services and optimal user experience.
[0110] Hence, instead of simply enlarging and up-scaling a selected
pixel area of an image which results in a loss of fidelity, the
client may simply use the SMF to efficiently switch to a desired
spatial representation. For example, a spatial representation
associated with a e.g. 960.times.640 resolution version of a source
stream may be selected, so that the top-left quarter is displayed
using a spatial segment stream of a 480.times.320 pixel resolution.
This way, the selected spatial representation may allow a user to
actually see more details when zooming in when compared with a
conventional digital zooming situation based on up-scaling existing
details in order to fit the entire display. Different spatial
representations in the SMF may allow a user to zoom to a particular
ROI without loosing fidelity and without requiring an excessive
amount of bandwidth.
[0111] FIG. 6A depicts a flow diagram of a process for delivering
spatially segmented content to a client according to an embodiment
of the invention. In this particular, embodiment, the SMF is stored
on the same delivery node DN as the spatial segment streams.
[0112] The process may start with the client selecting a particular
media file ashd31.asf (step 602) and sending a HTTP GET to
comprising a reference to the selected media file to a content
provider system (CPS) (in this case YouTube.TM.)(step 604). The CPS
may in response generate a redirect to the CDNCF of a CDN for
obtaining the SMF associated with the requested media file (steps
606 and 608). In response, the CDNCF may generate a redirect to a
content delivery node at which the requested SMF is stored (steps
609 and 610). The delivery node may then send the requested SMF to
the client.
[0113] Then as described above with reference to FIG. 5C, the
client may receive a ROI request from the user (step 612) and--in
response--the client may initiate a process for selecting a desired
spatial representation and for determining the spatial segment
subgroup, i.e. the spatial segment stream instances comprising the
selected ROI (step 614).
[0114] The client may set up streaming sessions, e.g. RTSP
streaming sessions, with the delivery node comprising the stored
SMF (steps 616 and 618). Upon reception of the requested spatial
segment streams (steps 620 and 622), the client may process the
spatial segment streams for construction of video images comprising
the selected ROI (step 624).
[0115] FIG. 6B depicts a flow diagram of client-controlled
spatially segmented content delivery according to a further
embodiment of the invention. In this case, the spatial segment
streams are stored on different delivery nodes D1 and D2. In that
case, the process may start in a similar way as in FIG. 6A. The
process may start with the client selecting a particular media file
ashd31.asf (step 630) and sending a HTTP GET comprising a reference
to the selected media file to a content provider system (CPS) (in
this case YouTube.TM.)(step 632). In response the CPS may generate
a redirect to the CDNCF of a CDN for obtaining the SMF associated
with the requested media file (steps 634 and 636). Thereafter, the
CDNCF may retrieve the SMF from the SMF storage associated with the
CDNCF. The SMF may have a format similar to the one described with
reference to FIG. 2E. In particular, for at least some of the
spatial segment streams, the SMF may define two or more delivery
nodes.
[0116] Hence, in that case the CDNCF will have to determine for
each spatial segment stream, which delivery nodes is most suitable
for transmitting the spatial segment stream to the client. For
example, based on the location of the client some delivery nodes
may be favorable over others. Moreover, some delivery nodes may be
avoided due to congestion problems or heavy processing loads. This
way, on the basis of the stored SMF, the CDNCF will generate a
"personal" SMF by selecting for a client for each spatial segment
stream the most suitable delivery nodes (step 638). This personal
SMF is sent to the client (step 640).
[0117] The client may receive a ROI request from the user (step
642). Then, on the basis of the personal SMF, the client may
initiate a process for selecting a desired spatial representation,
for determining the spatial segment subgroup and for determining
the locations of the delivery nodes for transmitting spatial
streams in the spatial segment subgroup to the client (step
644).
[0118] Thereafter, the client may set up streaming sessions, e.g.
RTSP streaming sessions, with the different identified delivery
nodes, in this case a first and second delivery node D1 and D2
respectively, for transmitting the spatial segment streams to the
client (steps 646-652). Upon reception of the requested spatial
segment streams (steps 650 and 652), the client may process the
spatial segment streams for construction of video images comprising
the selected ROI (step 654). Hence, in the process of FIG. 6B, the
client will receive an SMF, which is optimized in accordance with
specific client and CDN circumstances, e.g. client-delivery node
distance, processing load of the delivery nodes, etc.
[0119] While in FIGS. 6A and 6B the SMF retrieval is described on
the basis of HTTP, various other protocols such as SIP/SDP,
SIP/XML, HTTP/XML or SOAP may be used. Further, the request for
spatial segment streams may e.g. be based on RTSP/SDP, SIP/SDP,
SIP/XML, HTTP, IGMP. The segment delivery may be based on DVB-T,
DVB-H, RTP, HTTP (HAS) or UDP/RTP over IP-Multicast.
[0120] The SMF defining multiple spatial representations of a
source content stream may provides an efficient tool for ingestion
of content in a CDN and/or for content delivery from a CDN to a
client. It allows a client to efficiently determine spatial segment
streams associated with a ROI and to directly retrieve these
spatial segment streams for delivery nodes in a CDN. It further
allows an efficient way for a user to interact with content (e.g.
zooming and panning) without consuming excessive amounts of
bandwidth in the CDN and in the access lines connecting the clients
with the CDN. It is apparent that the a SMF allows for different
advantageous ways of controlling spatially segmented content.
[0121] FIG. 7A-7C depict at least part of SMF according to various
embodiments of the invention. FIG. 7A depicts part of an SMF
according to one embodiment of the invention. In particular, the
SMF in FIG. 7A defines the spatial position of a spatial segment
frame in an image frame to be displayed. This way for example the
left lower corner of spatial segment frames in a particular spatial
segment stream may be related to an absolute display position
1000,500.
[0122] The use of such coordinate system allows segment frames in
neighboring spatial segment streams to overlap thereby enabling
improved panning and zooming methods. For example, an overlap may
provide a smoother panning operation when switching to a
neighboring spatial segment stream
[0123] FIG. 7B depicts part of an SMF according to an embodiment of
the invention comprising inter-representation information. When
zooming within spatially segmented content comprising a large
number of spatial representations each defining multiple spatial
segment, it may be computationally expensive to find the desired
one or more references to spatial segment streams associated with a
particular ROI and a resolution version in another spatial
representation.
[0124] For these situations it may be useful to configure the SMF
to also comprise information about the correlation between
different spatial representations wherein each spatial
representation is associated with a particular resolution version
of the spatially segmented video content. For example, the spatial
segment stream at display position 1000,500 in the first spatial
representation is related to a spatial segment stream at display
position 500,250 in a second spatial representation and to a
spatial segment stream at display position 250,125 in a third
spatial representation.
[0125] The use of inter-representation information may be
especially advantageous when combined with an absolute pixel
coordinate system as described with reference to FIG. 7A. The
inter-representation may for example simplify the matching of a ROI
with the appropriate spatial representation during zoom
operations.
[0126] FIG. 7C depicts part of an SMF according to an embodiment of
the invention wherein normalized pixel coordinates is selected as
the center of the video image frame) define an absolute position
within an image frame associated with, a spatial representation
independent of the selected resolution version. Such normalized
coordinate system may range from 0 to 1 using (0.5,0.5) as the
origin. The use of such normalized coordinate system allows simpler
switching between different spatial representations within a SMF
during zooming operations.
[0127] FIGS. 8A and 8B further illustrate the use of different
coordinate systems for defining the position of spatial segment
frames in an image frame. FIG. 8A depicts several different
coordinate systems, i.e. a positioning system based on rows and
columns 802, an absolute coordinate system 804 and a normalized
coordinate system 806 for positioning a segment frame in an image
frame. FIG. 8B depicts a single spatial representation comprising a
matrix of spatial segment stream instances, whereby neighboring
spatial segment frames overlap (e.g. 50% overlap).
[0128] In order to process and display content on the basis of
individually transmitted spatial segment streams originating from
different delivery nodes, specific functionality needs to be
implemented in the client. In particular, frame level
synchronization of spatial segment streams originating from
different transmission nodes (such as the delivery nodes of a CDN)
is needed so that the individual spatial segment frames may be
stitched into a single image frame for display, wherein the image
frame comprises the ROI.
[0129] Synchronization is required as the data exchange between the
client and the delivery nodes is subjected to unknown delays in the
network of various origins e.g. transmission delays, differences in
network routes and differences in coding and decoding delays. As a
consequence, requests may arrive at different delivery nodes at
different times leading to different starting times of transmission
of the spatial segment streams associated with one to be produced
video.
[0130] Mavlankar et al. uses multiple instances of a media player
for displaying tiled video streams simultaneously in a synchronized
manner. Such solution however is not suitable for many low-power
media processing devices, such as smart phones, tablets, etc. These
devices generally allow only a single decoder instance running at
the same time. Moreover, Mavlankar does not address the problem of
synchronizing multiple segment streams originating from different
delivery nodes. In particular, Mavlankar does not addresses the
problem that between the request of new spatial segment streams and
the reception of the spatially segment streams a delay is present
and that this delay may vary between different delivery nodes in
the CDN.
[0131] Buffering and synchronization issues may be avoided or at
least significant reduced by encapsulating spatial segment streams
in a single transport container. However, transport of the spatial
segment streams in a single container significantly reduces the
flexibility of streaming of spatially segmented content since it
does not allow individual spatial segments streams to be directly
streamed from different delivery nodes to a client. Hence, in order
to maintain the advantages of the proposed method for spatially
segmented content streaming synchronization of the spatial segment
streams at the client is required.
[0132] FIG. 9 depicts a schematic of a media engine 902 configured
for synchronizing, decoding and spatial aligning spatial segment
frames according to various embodiments of the invention. The media
engine may be included in a client as described with reference to
FIG. 5A.
[0133] The media engine 902 may comprise a buffer 903 comprising
multiple adaptive buffer instances for buffering multiple
packetized spatial segment streams 906.sub.1-4. Using
synchronization information, the synchronization engine 910 may
control the buffer in order to synchronize packets in the spatial
segment streams.
[0134] Synchronization information may comprise timing information
in the headers of the packets in a spatial segment stream. Such
timing information may include time stamps, e.g. RTP time stamps,
indicating the relative position of an image frame in a stream.
Timing information may also relate to RTP time stamp information
exchanged between an RTSP server and a client. In further
embodiment, timing information may relate to time segments, e.g.
DASH time segments specified in a HTTP DASH manifest file and time
stamped frames (e.g. MPEG) or sequentially numbered frames (e.g.
AVI) contained in such DASH time segment.
[0135] As already described with reference to FIG. 4,
synchronization information 911, e.g. an RTP offset value, may also
be included in the SMF 913 and sent by the control engine 915 to
the synchronization module. In another embodiment, synchronization
information 917 may be sent in an in-band or out-of-band channel to
the synchronization engine.
[0136] The timing information may be used to control the one or
more adaptive buffer instances in the buffer module. For example,
on the basis of timing information of different spatial segment
streams time delays for the adaptive buffer instances may be
determined such that the output of the buffer may generate
time-aligned (i.e. time-synchronized) packets.
[0137] Synchronized packets at the output of the buffer may be
multiplexed by a multiplexer (not shown) into a stream of
time-synchronized packets 912 comprising frames which are logically
arrange in so-called groups of pictures (GOPs) so that a decoder
916 may sequentially decode the packets into a stream of decoded
spatial segment frames 920. In a further embodiment, the media
engine may comprise two or more decoders 920 for decoding data
originating from the buffer.
[0138] In one embodiment, the decoded segment frames may be ordered
in a sequence of spatial segment subgroups 921.sub.1-4. A stitching
engine may stitch the spatial segment frames associated with each
spatial segment subgroup into a video image frame comprising the
ROI. The stitching engine may use e.g. the position coordinates in
the SMF, in order to spatially align the spatial segment frames in
the spatial segment subgroup into a seamless video image frame 922
for display to a user.
[0139] The process of image stitching encompasses the technique of
reproducing a video image on the basis of spatial segments frames
by spatially aligning the spatial segment frames of a spatial
segment subgroup. In further embodiments, image-processing
techniques like image registration, calibration and blending may be
used to improve the stitching process. Image registration may
involve matching of image features between two neighbouring spatial
segment frames. Direct alignment methods may be used to search for
image alignments that minimize the sum of absolute differences
between overlapping pixels. Image calibration may involve
minimizing optical differences and/or distortions between images
and blending may involve adjusting colours between images to
compensate for exposure differences.
[0140] FIGS. 10A and 10B illustrate synchronized and stitched
spatial segment frames based on different streaming protocols. In
particular, FIG. 10A depicts spatial segment frames (in this case
four) which are synchronized and stitched into one image frame
wherein the spatial segment frames are transported to the client
using an adaptive streaming protocol, e.g. the HTTP adaptive stream
(HAS) protocol as defined in MPEG DASH ISO-IEC_23001-6 v12.
[0141] Each HAS-type spatial segment stream may be divided into N
equally sized temporal segments (hereafter referred to as
spatio-temporal segments or ST segments). Each ST segment is
associated with a particular period, e.g. 2-4 seconds, in a spatial
segment stream. The position of a ST segment in the spatial segment
stream is determined by a monotonically increasing (or decreasing)
segment number.
[0142] Each ST segment further comprises a predetermined number of
sequential spatial segment frames, wherein the position of a
spatial segment frame in a ST segment may be determined by a
monotonically increasing time stamp or frame number (depending on
the used transport container). For example, each numbered ST
segment may comprise a predetermined number of monotonically
numbered or time-stamped spatial segment frames. Hence, the
(temporal) position of each spatial segment frame in a HAS-type
spatial segment stream (or ST segment stream) is uniquely
determined by a segment number and a frame number within that ST
segment.
[0143] The division of a spatial segment stream in ST segments of
equal length may be facilitated by using the video encoding
parameters, such as the length and structure of the Group of
Pictures (GOP), video frame rate and overall bit rate for each of
the individual spatial segment streams.
[0144] The temporal format of a HAS-type spatial segment stream,
e.g. the time period associated with each TS segment, the number of
frames in a TS segment, etc., may be described in a HAS manifest
file (see FIG. 3 wherein a HTTP spatial segment stream is defined
in accordance with an HAS manifest file). Hence, in HAS-type
spatially segmented content the SMF may define the spatial format
of the content and may refer to a further manifest file, the HAS
manifest file, for defining the temporal format of the content.
[0145] Synchronisation and stitching of spatial segment frames in
ST segmented contend may be realized on the basis of the segment
number and the frame number as illustrated in FIG. 10A For example,
if the buffer in the media engine receives four separate ST segment
streams 1002.sub.1-4, the synchronization engine may process the ST
segment streams in accordance with a segment number (the time
segment number N in FIG. 10A) and frame number (the segment frame
number #M in FIG. 10A). For example, the synchronization engine may
determine for each ST segment stream associated with matrix
positions (1,1)(1,2)(2,1) and (2,2), a spatial segment frame
associated with a particular segment number and frame number is
selected by the synchronization engine and send to the output of
the buffer.
[0146] The encoded spatial segment frames at the output of the
buffer may be decoded and the decoded spatial segment frames may be
stitched into a video image for display in a similar way as
described with reference to FIG. 9.
[0147] FIG. 10B depicts spatial segment frames (in this example
four) which are synchronized and stitched into one image frame
wherein the spatial segment frames are transported to the client
using the RTP/RTSP streaming protocol. In that case, the buffer may
receive four RTP streams comprising time-stamped RTP packets. The
value of the RTP time stamp increase monotonically and may start
with a random RTP offset value. Hence, in order to determine the
relation between spatial segment frames in different RTP-type
spatial segment stream, knowledge of the RTP offset values is
required.
[0148] As will be described hereunder in more detail, when
requesting an RTSP server for a specific spatial segment stream,
the RTSP server may signal RTP time stamp information to the
client. On the basis of this RTP time stamp information the random
RTP offset value may be determined. Alternatively, when requesting
a specific spatial segment stream, the request may comprise a
particular predetermined RTP offset value, which the RTSP server
will use for streaming to the client. Such predetermined RTP offset
value may be defined in the SMF (see for example FIG. 3).
[0149] Hence, the synchronization engine may receive the RTP offset
values associated with the buffered spatial segment streams. These
RTP offset values may be determined in the SMF 1006. Using the RTP
offset values the synchronization engine may then determine the
relation between RTP packets in each of the spatial segment stream.
For example, in FIG. 10B, the synchronization engine may receive
four RTP offset values associated with four different spatial
segment streams 1004.sub.1-4. Using these offset values, the
synchronization engine may then determine packets in each of the
buffered spatial segment stream. For example, an RTP packet with
RTP timestamp 8000 (and RTP offset 4000) in spatial segment stream
(1,1) belongs to an RTP packet in spatial segment stream (1,2) with
RTP timestamp 6000 and RTP offset 2000, to an RTP packet in spatial
segment stream (2,1) with RTP time stamp 11000 and RTP offset 7000,
and an RTP packet in spatial segment stream (2,2) with RTP time
stamp 5000 and RTP offset value 1000.
[0150] The synchronization engine may send these packets to the
output of the buffer. Where after the packets are decoded into
spatial segment frames and stitched into a video image for display
in a similar way as described with reference to FIG. 9.
[0151] The above described synchronization process allows
frame-level synchronization of the spatial image frames which is
required in order to construct seamless video images on the basis
of two or more spatial segment streams.
[0152] Other known synchronization schemes such as the use of
so-called RTCP Sender Reports, which rely on the less accurate NTP
clock information are not suitable for synchronization of spatial
segment streams.
[0153] FIG. 11 depicts a process 1100 for determining
synchronization information for synchronizing spatially segment
content according to an embodiment of the invention. In particular,
FIG. 11 depicts a process for determining RTP offset values
associated with different spatial segment streams. The process may
start with a client sending an RTSP PLAY request to a first
delivery node DN1, requesting the RTSP server to start streaming a
first spatial segment stream stream1_1RTSP to the client starting.
The range parameter in the RTSP PLAY request instructs the RTSP
server to start streaming at position t=10 seconds in the stream
(step 1102). In response, DN 1 sends an RTSP 200 OK response to the
client confirming that it execute the request. Moreover, the
RTP-info field in the response indicates that the first packet of
the stream will have RTP time stamp 5000 (step 1104). Then, on the
basis of the streaming clock rate of 100 frames per second, the
client may calculate an expected RTP time stamp of 1000 (i.e.
100.times.10 second). The RTP offset value of 4000 may then be
determined by subtracting the received RTP time stamp from the
expected RTP timestamp (step 1106).
[0154] In a similar way, at t=20 seconds the client may then
request a further spatial segment stream stream 1_2 from a second
delivery node DN2 (step 1108). In response, DN2 sends an RTSP 200
OK message to the client indicating that the first RTP packet will
have RTP time stamp 7000 (step 1110). Using this information, the
client may then determine that the RTP timestamp offset for the
second spatial segment stream is 5000 (step 1112). Hence, on the
basis of RTP timing information in the RTSP signalling between the
client and the delivery nodes, the client may efficiently determine
the RTP offset values associated with each spatial segment
stream.
[0155] From the above, it follows that the SMF allows a client to
directly access different delivery nodes in a CDN so that fast
content delivery is achieved. Fast content delivery is required in
order to allow good user-interaction, e.g. zooming and panning,
with displayed spatially segmented content. Some delay between
request and reception of content is however inevitable. These
delays may also differ between different delivery nodes in a CDN.
FIG. 12 depicts an example of a client requesting content from two
different delivery nodes DN1 and DN2.
[0156] At t=0 the client may send an RTSP PLAY request to DN1 (step
1202). In response to the request, DN1 may start sending RTP
packets to the client (step 1204). Due to some delays, RTP packet
associated with t=0 will be received by the client at t=1.
Similarly, the RTP packet transmitted at t=1 will be received by
the client at t=2 and the RTP packet transmitted at t=2 will be
received by the client at t=3. At that moment the client may decide
to start play-out of the first RTP packet t=0 (step 1206), while
the other two RTP packets t=1 and 2 are still buffered.
[0157] Then, while watching, play-out of RTP packet t=1, the user
may decided at t=4 to request a panning operation. In response to
this request, the client may request streaming of a predetermined
spatial segment stream from a second delivery node DN2 starting
from t=4 (i.e. range parameter is set to t=4)(step 1208). DN2
however has a slower response time, so that it sends the RTP packet
t=4 at t=6 to the client (step 1210). Due to this delay, the client
is only capable of displaying the spatial segment stream
originating from DN2 at t=7 (step 1212), 3 seconds after the user
has initiated the panning request. Hence, schemes are desired to
reduce the effect of the delay as much as possible.
[0158] FIG. 13 depicts a process for improving the user-interaction
with the spatially segmented content according to an embodiment of
the invention. In particular, FIG. 13 depicts a client and two
delivery nodes similar to those described with reference to FIG. 12
(i.e. a first delivery node DN1 with a response of 1 second and a
slower second delivery node DN2 with a response delay of 2
seconds).
[0159] After requesting DN1 to start streaming a first spatial
segment stream from starting position t=0 (step 1302), the client
may start play-out of the RTP packets at t=3 (i.e. after having
buffered two RTP packets)(step 1304). Then, while watching play-out
of RTP packet t=1, the user may decide at t=4 to request a panning
operation. In response to this request, the client may send two
RTSP PLAY request to the second delivery node DN2: a first RTSP
PLAY request requesting the second delivery node to start streaming
the second spatial segment stream starting from position t=4s
(using the range parameter in the RTSP PLAY request) (step 1306);
and, a second RTSP PLAY request requesting DN2 to start streaming
the second spatial segment stream from position t=1 to position t=4
at a faster-than-real time streaming rate (using the range
parameter and a SPEED parameter in the RTSP PLAY request) (step
1308).
[0160] In response, (at t=6) the second delivery node may start
sending RTP packets t=1,2,3 and 4 at an increased rate to the
client (step 1310) so that at t=6 the client may start play-out of
RTP packet t=3 from the second delivery node (step 1312). At the
same time, the first RTSP PLAY request may initiate transmission of
RTP packets at a normal rate so that once the client play-out has
arrived at t=4, the delivery of RTP packets at an increased rate
may be terminated. This way, the user will notice the panning
operation with a delay of two second, thus decreasing effects in
the user-interaction due to network delays.
[0161] A temporarily delivery of RTP packets at an increased rate
may be used in Video-on-Demand situations. For life-streaming
situations, which do not allow faster-than-realtime streaming,
another solution may be used. In that case, fast-channel change
techniques based on IP-Multicast as described in the IEFT document
"Unicast-based rapid acquisition of multicast RTP sessions"
http://tools.ieff.org/html/draft-ieff-avt-rapid-acquisition-for-rtp-17
may be used.
[0162] It is to be understood that any feature described in
relation to any one embodiment may be used alone, or in combination
with other features described, and may also be used in combination
with one or more features of any other of the embodiments, or any
combination of any other of the embodiments.
[0163] One embodiment of the invention may be implemented as a
program product for use with a computer system. The program(s) of
the program product define functions of the embodiments (including
the methods described herein) and can be contained on a variety of
computer-readable storage media. Illustrative computer-readable
storage media include, but are not limited to: (i) non-writable
storage media (e.g., read-only memory devices within a computer
such as CD-ROM disks readable by a CD-ROM drive, flash memory, ROM
chips or any type of solid-state non-volatile semiconductor memory)
on which information is permanently stored; and (ii) writable
storage media (e.g., floppy disks within a diskette drive or
hard-disk drive or any type of solid-state random-access
semiconductor memory) on which alterable information is stored. The
invention is not limited to the embodiments described above, which
may be varied within the scope of the accompanying claims.
* * * * *
References