U.S. patent number RE36,822 [Application Number 09/165,356] was granted by the patent office on 2000-08-15 for moving image signal coding apparatus and coded signal decoding apparatus.
This patent grant is currently assigned to Victor Company of Japan, Ltd.. Invention is credited to Kenji Sugiyama.
United States Patent |
RE36,822 |
Sugiyama |
August 15, 2000 |
Moving image signal coding apparatus and coded signal decoding
apparatus
Abstract
A coding apparatus which codes moving image signals into block
units, is configured from a signal processing element which
performs motion compensation for moving image signals for over a
plural number of frames or fields and codes inter-image signals,
and a transfer element which recombines coded information for each
block coded by said processing element, into macroblock units which
are a plural number of block units of each type of coded
information, and transfers them. In addition, a decoding apparatus
for moving image signals which have been coded in block units is
configured from a detector element which detects transfer code
errors for each type of coded information, and a processing element
which performs motion compensation and inter-image processing of
the coded information using only correct frames which do not
include transfer code errors, and without using frames which have
transfer code errors, by changing a method of inter-frame
processing for motion compensation in accordance with the transfer
coding errors in the coded information which has been detected for
each type.
Inventors: |
Sugiyama; Kenji (Yokosuka,
JP) |
Assignee: |
Victor Company of Japan, Ltd.
(Yokohama, JP)
|
Family
ID: |
18131790 |
Appl.
No.: |
09/165,356 |
Filed: |
October 2, 1998 |
Related U.S. Patent Documents
|
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
Issue Date |
|
|
324481 |
Oct 18, 1994 |
|
|
|
|
972564 |
Nov 6, 1992 |
|
|
|
Reissue of: |
666687 |
Jan 17, 1996 |
05748784 |
May 5, 1998 |
|
|
Foreign Application Priority Data
|
|
|
|
|
Nov 8, 1991 [JP] |
|
|
3-321368 |
|
Current U.S.
Class: |
382/236;
375/240.16; 375/240.27 |
Current CPC
Class: |
H04N
19/51 (20141101); H04N 19/577 (20141101); H04N
19/517 (20141101); H04N 19/89 (20141101) |
Current International
Class: |
H04N
7/46 (20060101); G06K 009/36 () |
Field of
Search: |
;382/232,234,236,238,250
;358/342 ;348/409,415,416,420,421,423,384 |
References Cited
[Referenced By]
U.S. Patent Documents
Other References
Verbiest et al. "The impact of the ATM Concept on Video coding",
Dec. 1998, pp. 1623-1632 IEEE Journal on Selected Areas in
Communications..
|
Primary Examiner: Couso; Jose L.
Attorney, Agent or Firm: Jacobson, Price, Holman &
Stern, PLLC
Parent Case Text
This is a continuation of application Ser. No. 08/324,481, filed
Oct. 18, 1994 which was abandoned upon the filing hereof, and
which, in turn, is a continuation of application Ser. No.
07/972,564, filed Nov. 6, 1992 now abandoned.
Claims
What is claimed is:
1. A coding apparatus for coding moving image signals into block
units, comprising:
image processing means for performing motion compensation between a
plural number of frames.Iadd./fields per .Iaddend..[.in each.].
block of a plurality of blocks which constitute one .[.image
screen.]. .Iadd.frame/field.Iaddend., thereby outputting motion
vector data, inter-image processing data indicative of what
inter-image processing is performed, and image data,
respectively;
a plurality of data memory means for storing multiple types of said
motion vector data, said inter-image processing data, and said
image data; and
transfer means for time division multiplexing said multiple types
of said motion vector data, said inter-image processing data, and
said image data, such that said multiple types of said motion
vector data for a group of blocks are time multiplexed together in
one time division, the inter-image processing data for said group
of blocks is time multiplexed together in another time division,
and said image data for said group of blocks is time multiplexed to
a further time division, and thereby transferring the multiplexed
data.
2. The coding apparatus of claim 1, wherein:
said image processing means comprises:
a first changeover switch which switches signals of a bidirection
(B) frame.Iadd./field .Iaddend.predicted for front and back of
image signals supplied via the first changeover switch and an image
signal input terminal, and both signals of a skip-predicted
prediction (P) frame.Iadd./field .Iaddend.and an
independently-coded intra- (I) frame/.Iadd.field.Iaddend.;
a first frame.Iadd./field .Iaddend.memory which stores signals of
said B-frame.Iadd./field .Iaddend.so as to delay them until the end
of coding of signals of both said I-frame.Iadd./field .Iaddend.and
P-frame.Iadd./field.Iaddend.;
a second changeover switch which switches B-frame.Iadd./field
.Iaddend.signals from said first frame.Iadd./field .Iaddend.memory
and said I- and P-frame.Iadd./field .Iaddend.signals switched by
said first changeover switch;
first and second motion vector estimators which estimate motion
vectors of signals of both said I-frame.Iadd./field .Iaddend.and
said P-frame.Iadd./field.Iaddend.;
an adaptive predictor which receives signals of frames.Iadd./fields
.Iaddend.switched by said second changeover switch;
a residual subtracter which calculates a remainder of signals of
frames.Iadd./fields .Iaddend.switched by said second changeover
switch, and prediction mode of output information output from said
adaptive predictor;
an intra-frame.Iadd./field .Iaddend.coder which codes signals
output from said residual subtracter;
a third changeover switch which switches coded signals from said
intra-frame.Iadd./field .Iaddend.coder on the basis of
B-frames.Iadd./fields.Iaddend., and I-frames.Iadd./fields
.Iaddend.and P-frames.Iadd./fields.Iaddend.;
an intra-frame.Iadd./field .Iaddend.decoder which decodes coded
signals of each frame.Iadd./field .Iaddend.and from said third
changeover switch;
a residual adder which adds signals from said
intra-frame.Iadd./field .Iaddend.decoder and said adaptive
predictor;
a second frame.Iadd./field .Iaddend.memory which stores one of
I-frame.Iadd./field .Iaddend.and P-frame.Iadd./field
.Iaddend.signals from said adder so as to delay them;
a third frame.Iadd./field .Iaddend.memory which stores another of
I-frame.Iadd./field .Iaddend.and P-frame.Iadd./field
.Iaddend.signals which have passed through said second
frame.Iadd./field .Iaddend.memory;
said first motion vector estimator which estimates a motion vector
of moving image signals for one of I-frames.Iadd./fields
.Iaddend.and P-frames.Iadd./fields .Iaddend.said motion vector
being supplied via said first changeover switch;
said second motion vector estimator which estimates .[.an other.].
.Iadd.another .Iaddend.motion vector of moving image signals for
another of I-frames.Iadd./fields .Iaddend.and
P-frames.Iadd./fields.Iaddend., said other motion vector being
supplied via said first changeover switch;
a first motion compensator which performs motion compensation of
moving image signals of F-frames.Iadd./fields.Iaddend., by an
output of said third frame.Iadd./field .Iaddend.memory and an
output of said first motion vector estimator; and
a second motion compensator which performs motion compensation of
moving image signals of B-frames.Iadd./fields.Iaddend., by an
output of said second frame.Iadd./field .Iaddend.memory and an
output of said second motion vector estimator; and wherein
said adaptive predictor uses both reproduced image signals which
have been moved by a motion vector portion of F-frames.Iadd./fields
.Iaddend.and B-frames.Iadd./fields .Iaddend.supplied from said
first and second motion compensators and moving image signals
supplied via said second changeover switch, as the basis for
creating a plural number of prediction signals from a plural number
of signals which have been motion compensation by the same clock
and for which motion vector detection has been performed, and
outputs an optimum prediction signal within said plural number of
prediction signals as prediction mode signals to said residual
subtracter, said residual adder and said transfer means.
3. The coding apparatus according to claim 2, wherein said
plurality of data memory means comprises;
a first memory which stores coded signals output from said
intra-frame.Iadd./field .Iaddend.coder of said processing
means;
a second memory which stores prediction mode signals output from
said adaptive predictor;
a third memory which stores first motion vector signals output from
said first motion vector estimator; and
a fourth memory which stores second motion vector signals output
from said second motion vector estimator; and
wherein said transfer means comprises a coder which successively
selects and outputs signals stored in said first through fourth
memories in response to a required number of block pulses.
4. The coding apparatus according to claim 1, wherein said
plurality of data memory means comprises:
a first memory which stores coded signals output from said image
processing means;
a second memory which stores prediction mode signals output from
said image processing means;
a third memory which stores the first motion vector signals output
from said image processing means; and
a fourth memory which stores second motion vector signals output
from said image processing means; and
wherein said transfer means comprises a selector which successively
selects and outputs signals stored in said first through fourth
memories in response to a required number of clock pulses.
5. A decoding apparatus for decoding moving image signals coded in
block units, comprising:
detection means for receiving input data which includes at least
motion vector data, inter-image processing data indicative of what
inter-image processing is performed, and image data which are
time-division multiplexed and received by the detection means, said
detecting means detecting transfer code errors for each of said
inter-image processing data and said image data, and outputting the
coded information for each data type having said code errors;
and
processing means for performing motion compensation and inter-image
processing of said coded information using only frames.Iadd./fields
.Iaddend.which do not include said transfer code error within a
plurality of frames.Iadd./fields .Iaddend.which are to be used for
prediction purposes, and without the use of frames.Iadd./fields
.Iaddend.which have said transfer code errors within said plurality
of frames.Iadd./fields .Iaddend.which are to be used for prediction
purposes, by selecting a method of inter-frame.Iadd./field
.Iaddend.processing for motion compensation in accordance with said
detected transfer code errors, wherein one .[.image screen.].
.Iadd.frame/field .Iaddend.comprises a plurality of blocks, .[.each
block comprises a plurality of frames,.]. and wherein motion
compensation and inter-image processing is performed .[.in each.].
.Iadd.per .Iaddend.block.
6. The decoding apparatus according to claim 5, wherein said
processing means uses only those frames.Iadd./fields .Iaddend.which
do not have coding errors, to perform motion compensation of said
coded information.
7. The decoding apparatus according to claim 5, wherein said
detection means comprises an error detector which detects said
transfer code errors included in coded information supplied from a
coding apparatus via a data input terminal.
8. The decoding apparatus according to claim 5, wherein said
processing means comprises:
an adaptive predictor responsive to output of said detection means
for generating as output prediction signals from prediction mode
information included in said coded information; and
a variable adder responsive to output of said detection means and
said adaptive predictor for adding decoded signals of said coded
information and prediction signals of said adaptive predictor.
9. The decoding apparatus according to claim 5, and further
comprising:
a selector which separates DCT (discrete cosine transform)
information, prediction mode information, first motion vector
information and second motion vector information multiplexed in
said coded information signals;
first, second, third and fourth memories which respectively store
said DCT information, prediction mode information, first motion
vector information and second motion vector information separated
by selector; and
wherein said detecting means comprises:
an error detector which detects transfer code errors in said coded
information signals;
an intra-frame.Iadd./field .Iaddend.decoder which decodes said DCT
information stored in said first memory;
a first frame.Iadd./field .Iaddend.memory;
a second frame.Iadd./field .Iaddend.memory;
a first motion compensator which uses signals of the first
frame.Iadd./field .Iaddend.memory as the basis for performing
motion compensation for said first motion vector information stored
in said third memory;
a second motion compensator which uses signals of the second
frame.Iadd./field .Iaddend.memory as the basis for performing
motion compensation for said second motion vector information
stored in said fourth memory;
an adaptive predictor which uses first and second compensation
signals output from said first and second motion compensator, and
output signals of said error detector as the basis for generating
prediction signals from said prediction mode information stored in
said second memory;
a variable adder which adds prediction signals from said adaptive
predictor, decoded signals from said intra-frame.Iadd./field
.Iaddend.decoder and output signals of said error detector; and
changeover switch means for switching between outputs of said
variable adder and which is connected to said first and second
frame.Iadd./field .Iaddend.memory.
10. The decoder apparatus according to claim 9, further provided
with a matching decider which judges matching between decoded
signals from said intra-frame.Iadd./field .Iaddend.decoder and
prediction signals from said adaptive predictor and outputs to said
variable adder.
11. A coding apparatus for coding moving image signals into block
units, comprising:
image processing means for dividing one .[.screen.].
.Iadd.frame/field .Iaddend.into a plurality of blocks, .[.each
block comprising a plurality of frames,.]. and performing motion
compensation image processing between a plural number of
frames.Iadd./fields .Iaddend..[.in each.]. .Iadd.per
.Iaddend.block, and outputting primary motion vector data which is
used for a motion compensation, .[.in said primary motion vector
data, in.]. inter-image processing data indicative of what
inter-image processing is performed, and .[.in.]. image data
.Iadd.which is inter-image processed.Iaddend.;
motion vector detection means for detecting secondary motion vector
data which is used in providing error concealment in a decoding
apparatus when code errors occur during a data transmission;
and
transfer means for multiplexing said primary motion vector data,
said secondary motion vector data, said inter-image processing data
and said image data.
12. A decoding apparatus for decoding moving image signals coded in
block units, comprising:
detection means for receiving input data including at least primary
motion vector data which is used for motion compensation,
inter-image processing data indicative of what inter-image
processing is performed, and image data .Iadd.which is inter-image
processed.Iaddend., and for detecting code errors included in said
input data;
reception means for receiving secondary motion vector data which is
used in providing error concealment .[.in said decoding
apparatus.]. only when code errors occur during a data
transmission; and
image processing means for performing motion compensation and
inter-image processing between a plurality of frames.Iadd./fields
.Iaddend..[.that make up a block, and wherein.]. .Iadd.per block of
.Iaddend.a plurality of blocks .[.make up an image screen.].
.Iadd.which constitutes one frame/field.Iaddend., the motion
compensation and inter-image processing employing said primary
motion vector data only when there is no code error in said input
data, and said image processing means performing motion
compensation and inter-image processing using said secondary motion
vector data when the input data has code errors.
13. A decoding apparatus for decoding moving image signals coded in
block units, comprising:
detection means for receiving input data which includes at least
motion vector data, inter-image processing data, and image data and
which is transferred by time division multiplexing said input data
such that said motion vector data for a group of blocks are time
multiplexed together in one time division, the inter-image
processing data for said group of blocks is time multiplexed
together in another time division, and said image data for said
group of blocks is time multiplexed to a further time division,
said detecting means detecting transfer code errors for each of
said inter-image processing data and said image data, and
outputting coded information signals for each type having said code
errors:
wherein said detection means comprises
an error detector which detects transfer cod errors in said coded
information signals;
and intra-frame.Iadd./field .Iaddend.decoder which decodes a
discrete cosine transform (DCT) information in a first memory;
a predictor which uses first and second compensation signals output
from a first and second motion compensators, respectively, and
output signals of said error detector as the basis for generating
prediction signals from prediction mode information stored in a
second memory; and
a variable adder which adds said prediction signals from said
predictor, decoded signals from said intra-frame.Iadd./field
.Iaddend.decoder and output signals from said error detector.
Description
BACKGROUND OF THE INVENTION
The present invention relates to high-efficiency coding and
decoding apparatus which are used in recording, transfer and
display apparatus which perform digital signal processing, and
which perform efficient coding and decoding, and in particular, to
coding and decoding apparatus which perform inter-image processing
of moving image signals have small deterioration of image quality
even when there are transmission errors.
High-efficiency coding of moving image signals (moving images) can
involve interframe predictive coding which uses the correlation
between frames of image signals and uses a frame for which coding
has been performed, to predict and code only the prediction error.
In recent years, motion compensation predictive coding has become
the general method used for prediction in accordance with motion of
an image.
On the other hand, in coding in which a storage media is the
object, intraframe independent coding is performed without
interframe prediction for each of several frames and this enables
random access and high-speed search.
In addition, there is also known a method such as MPEG (ISO-IEC)
which uses skip prediction and pre- and post-prediction between
skip predictions to raise the coding efficiency. With the MPEG
method, differences in the method of prediction mean that a frame
can be an I (intra) frame coded independently within a frame, a P
(Prediction) frame which is skip predicted, or a B (Bi-directional)
frame which is pre- and post-predicted.
The following is a description of a detailed configuration of a
conventional coding apparatus.
FIG. 1 shows an example of the configuration of a coding apparatus
of the MPEG type. Here, the frame types of I, P and B cause the
changeover switch 2, 4, 22 to be controlled by sync signals
separated from input signals, and to be switched to the positions
shown in the figure.
Image signals input from an image input terminal 1 are directly led
to a predictive, subtracter 5 via the changeover switches 2 and 4
in the case of I or P frames, while B frames are led to the
predictive subtracter 5 after having been delayed until there is
pre- and post-I and P in a frame memory 3. In the predictive
subtracter 5, prediction signals arriving from an adaptive
predictor 42 are subtracted from input signals and a prediction
residual signal is output to become coded data compressed by coding
in an intraframe encoder 6.
In the intraframe encoder 6, DCT (discrete cosine transform) is
first preformed, and that conversion output is quantized, and given
a variable length coding such as Huffman coding or the like. That
compressed DCT information is applied to a multiplexer 40 and in
the case of I and P frames, is led to an intraframe decoder 21 via
the changeover switch 22.
The intraframe decoder 21 first decodes the variable length coding,
and replaces the fixed-length codes with quantized representative
values, and also performs reverse DCT to obtain the reproduced
signals. In the intraframe decoder 21, the reproduced prediction
error signals have the prediction signals added in a residual adder
20 to produce the reproduced image signals. The reproduced image
signals are stored in a frame memory 19 while the signal that have
been stored in the frame memory 19 until that time are transferred
to a frame memory 18.
The output of the frame memory 19 is given to a motion compensator
15 and a motion vector detector 17, and the output of the frame
memory 18 is given to a motion compensator 14 and the motion vector
detector 16.
For each block of 16.times.16 picture elements, the motion vector
estimators 16 and 17 detect the motion vectors between the input
signals and the signals given to the frame memories 18 and 19. The
motion vector information is given to the motion compensators 14
and 15 and also to a multiplexer 40. The motion compensators 14 and
15 spatially move reproduced image signals stored in the frame
memories 18 and 19 by the motion vector portion given from the
motion vector detector, and applies them to an adaptive predictor
42.
For the same block as the motion vector detection, the adaptive
predictor 42 creates four types of prediction signals from the two
signals (F and B) which have been motion compensated, and of those,
the optimum prediction signals is decided from matching with the
input signals which become the signals to be predicted.
The prediction mode used here is one of the four types of only the
"F" (Front: prediction signals from the frame temporally prior)
mode, only the "B" (Back: prediction signals from the frame
temporarily later) mode, the "(F+B)/2" mode or the "0" mode, with
the "0" mode being intraframe independent coding. The prediction
mode is only the "0" mode for I-frames, the "F" and "0" modes for
P-frames, or any of the four modes for B-frames.
The multiplexer 40 recombines the DCT information which is the
output of the intraframe encoder 6, the prediction mode information
(MODE) which is the output of the adaptive predictor 42, the motion
vector information (MVF and MVB) which is the output of the motion
vector detector 16, for each block (macroblock: MB) for which the
motion vector and the prediction mode have been determined, and
outputs them via a data output terminal 12, to the side of a
decoding apparatus. FIG. 6A shows the configuration of the data.
Here, there is no transfer of the motion vector information not
used in the prediction.
The following is a description of a conventional decoding
apparatus.
FIG. 2 is a view showing the configuration of a decoding apparatus.
Those portions which correspond to portions of the coding apparatus
of FIG. 1 are shown with corresponding numerals. The coded data
which is input from a data input terminal 30 is disassembled into
each information by a demultiplexer 41 and the DCT information is
applied to the intraframe decoder 21, the prediction mode
information is applied to an adaptive predictor 43, and the motion
vector information is applied to the motion compensators 14 and
15.
The DCT information is decoded by the intraframe decoder 21, and
prediction signals are added at the residual adder 20 to create the
reproduced image signals.
In the case of B-frames, reproduced image signals are immediately
outputted from a reproduced image signal output terminal 36 via
changeover switches 34 and 35, while I- and P-frames are stored in
the frame memory 19. The signals which have been stored in the
frame memory 19 up till that time are moved to the frame memory 18
and are outputted from the reproduced image signal output terminal
36 via the changeover switch 35.
The output of the frame memory 19 is applied to the motion
compensator 15 while the output of the frame memory 18 is applied
to the motion compensator 14. The motion compensators 14 and 15
spatially move the reproduction image signals stored in the frame
memory, by the motion vector portion given from the demultiplexer
41, and applies them to the adaptive predictor 43. The adaptive
predictor 43 makes the prediction signals from the prediction mode
information given from the demultiplexer 41 and outputs it to the
residual adder 20.
Here, the inter-image processing units are frames but the
description is the same if they are fields of interlace
signals.
When there is a coding error between the coding apparatus and its
decoding apparatus during transfer or recording, normal
demodulation does not occur and there is a deterioration in the
image quality. Coding errors result in cell loss in ATM
(asynchronous transfer mode) circuits when they occur in normal
circuits and recording media, and this loss in cell units becomes a
"dropout".
In this case, even for the case of the coding apparatus and the
decoding apparatus shown as the conventional example, there is
normally detection to the effect that a coding error has occurred
and so with prediction residual errors, the prediction residue is
not added and the reproduced image signals are the prediction
signals only, to result in there being no particularly large
deterioration. However, there is absolutely no decoding of a block
if there is "dropout" of the motion vector, adjacent blocks and the
like are used for interpolation within the same frame and there is
no image deterioration as a result.
On the other hand, with a coding method which periodically has
independent frames, the deterioration stops with the independent
frames and so this method appears advantageous at first. However,
coding errors in independent frames can only be compensated for
spatially and so there the deterioration becomes large, and the
image is influenced later. Furthermore, the amount of data for
independent frames is larger than that for prediction frames and so
when there are ten independent frames at once, the amount of data
is about 40% of the overall amount of data, and the influence of
coding errors becomes serious.
SUMMARY OF THE INVENTION
In order to solve the problems described above, the object of the
present invention is to provide a moving image coding apparatus and
decoding apparatus which always makes a plural number of frames
used in interimage processing and transfers information for the
motion compensation and inter-image processing method, which
detects errors by the decoding apparatus and switches to another
frame without the use of a frame which had an error in the
inter-image processing for each block, and which has no large image
deterioration even if there is a coding error in the transfer
path.
In order to attain this objective, as shown in FIG. 3, the present
invention is a moving image coding apparatus which codes moving
image signals in block units, and is a moving image coding
apparatus which has means (a predictive subtractor 5, an adaptive
predictor 13, and the like) for motion compensation inter-image
processing between a plural number of frames (or fields) and means
(memory 7, 8, 9 and 10 and selector 11) for combining coded
information for each block and which has been coded by the
processing means, for each type of coded information, into a plural
number of block units (a plural number of macroblock units) and for
then transferring it.
Furthermore, as shown in FIG. 4 for example, the present invention
is a moving image decoding apparatus which comprises a decoding
apparatus for moving image signals which have been coded in block
units, and has means (error detector 38) for detecting transfer
coding errors for each type of coded information, and means
(adaptive predictor 38, variable adder 33, and the like) for
performing motion compensation inter-image processing by changing
the method of motion compensation inter-image processing for each
block in accordance with errors in each type of coded information
which have been detected, and which either does not use decoded
signals of frames (or fields) having coding errors and limits use
to only decoded signals of correct frames (or fields), or
substitute them with decoded signals of other frames (or
fields).
A decoding apparatus always makes a plural number of frames for use
in the inter-image processing and recombines the information for
that motion compensation and inter-image processing method for each
type of information and so it is possible to lower the probability
that a plural number of pieces of information of the same block
will not be used even if error detection is performed in code units
of a certain quantity.
In the decoding apparatus, there is the detection of transfer
coding errors for each type of coded information and there is
switching to another frame instead of using signals of frames
having errors in the information for inter-image processing for
each block. There is therefore very little image deterioration.
In the moving image coding apparatus and decoding apparatus of the
present invention, the number of frames used for inter-image
processing is always made a plural number, and the information for
the motion compensation and inter-image processing method is
recombined into each type of information and transferred, with
errors for each type of information being detected and with
inter-image processing for each block being switched to another
frame without the use of signals of frames having errors in the
information, thereby lowering the probability that a plural number
of pieces of information in the same block will not be used even if
error detection is performed in code units of a certain amount, and
thereby enabling there to be little deterioration of the image
quality.
By this, it is possible to not have a large amount of deterioration
in the image quality even if there is a large number of errors in
the transfer path. Accordingly, coding errors are permissible and
it is not necessary to have a large amount of correction coding in
the transfer coding, with the result that the amount of data can be
reduced.
As has been described above, a moving image coding apparatus and
decoding apparatus of the present invention has advantageous
effects in its practical application.
BRIEF DESCRIPTION OF THE DRAWINGS
In the appended figures:
FIG. 1 is a block diagram showing an outline configuration of a
conventional moving image signal coding apparatus;
FIG. 2 is a block diagram showing an outline configuration of a
conventional moving image signal decoding apparatus;
FIG. 3 is a block diagram shown an outline configuration of a
moving image signal coding apparatus according to a first
embodiment of the present invention;
FIG. 4 is a block diagram showing an outline configuration of a
moving image signal decoding apparatus according to a first
embodiment of the present invention;
FIG. 5 is a block diagram showing an outline configuration of a
moving image signal decoding apparatus according to a second
embodiment of the present invention; and
FIGS. 6A and 6B are views showing the data configurations in a
conventional coding/decoding apparatus and the coding/decoding
apparatus according to the present invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
The following is a detailed description of the preferred
embodiments of the moving image coding apparatus and decoding
apparatus of the present invention, with reference to the appended
drawings.
First is a description of the coding apparatus of a first
embodiment of the present invention.
FIG. 3 is a block diagram showing an outline configuration of a
moving image signal coding apparatus according to the first
embodiment of the present invention. Those portions which
correspond to portions of the coding apparatus of FIG. 1 are shown
with corresponding numerals.
In FIG. 3, the coding processing is fundamentally the same, with
the operation of the changeover switches 2, 4 and 22, the
predictive subtracter 5, the intraframe encoder 6, the intraframe
decoder 21, the frame memories 18 and 19, the motion compensators
14 and 15, and the motion vector detectors 16 and 17 being the same
(as shown by the dots and the like where there is coding in block
units).
This coding apparatus differs from the conventional transfer method
(FIG. 2) for each of the information of the DCT information which
is the output of the intraframe encoder 6, the prediction mode
information (MODE) which is the output of the adaptive predictor
42, and the motion vector information (MVF and MVB) which is the
output of the motion vector estimators 16 and 17. More
specifically, memories 7, 8, 9 and 10 and the selector 11 are
configured so that each of the types of information is recombined
in the memories and then transferred.
The DCT information which is the output of the intraframe encoder
6, the prediction mode information (MODE) which is the output of an
adaptive predictor 13, and the motion vector information (MVF and
MVB) which is the output of the motion vector estimators 16 and 17
are all respectively stored once in the memories 7, 8, 19 and 10.
Then, at the time when from 30-300 macroblocks have been stored,
there is output via the sequential data output 12 selected by the
selector 11 and in the format shown in FIG. 6B. The size of the
plural number of block units (macroblocks) for each type of the
recombined coded information is sufficiently larger than the blocks
for error detection, and can be set to a size smaller than the
number of blocks of one frame. In this embodiment, dropout in cell
units in an ATM circuit is for several macroblocks and the number
of blocks of one frame is about 1350 macroblocks and so the size is
set to 30-300 macroblocks as described above.
In addition, the adaptive predictor 13 is conventionally used only
for the one frame (F-frames) for P frames but in the present
embodiment, the configuration enables its use for a plural number
of frames (F- and B-frames). Moreover, a motion vector is sent
without the use of the predictive mode because of error
correspondence in a decoder apparatus.
The following is a description of a decoder apparatus of the first
embodiment.
FIG. 4 is a block diagram showing an outline configuration of a
moving image signal decoding apparatus according to the first
embodiment of the present invention. Those portions which
correspond to portions of the coding apparatus of FIGS. 2 and 3 are
shown by the same numerals. In FIG. 4, the decoding processing is
fundamentally the same as that conventional example shown in FIG.
2, and the operation of the changeover switches 34 and 35, the
intraframe decoder 21, the frame memories 18 and 19 and the motion
compensators 14 and 15 are the same.
The differences with the conventional example (FIG. 2) is that
there are the memories 7, 8, 9 and 10, the selector 31, an error
detector 38 and a variable adder 33 which changes the gain of the
signals from the predictor and the intraframe decoding signals,
with there being a different method of handling of each piece of
information and different operation for an adaptive predictor 37
and the variable adder 33.
More specifically, the coded data signals are transferred by the
coding device and via the data input terminal 30 and arrive at the
selector 31 where they are separated into multiplexer DCT
information, prediction mode information and motion vector
information which are respectively stored in the memories 7, 8, 9
and 10.
Then, the DCT information which is stored in the memory 7 is
applied to the intraframe decoder 21, the prediction mode
information which is stored in the memory 8 is applied to the
adaptive predictor 37, and the motion vector information which is
stored in the memories 9 and 10 is applied to the motion
compensators 14 and 15.
On the other hand, the error detector 38 decodes the error
detection code of the transmission path coding generated by the
coding device, and judges the cell loss information for the ATM
circuit to decide if there is an error in what type of information
for the macroblock. The type of error changes the prediction mode
according to the rules shown in Table 1.
In Table 1, C is a current decoding signal which is the output of
the intraframe decoder. When there is dropout of mode information,
the frame is which of B or F is temporally closer to C.
TABLE 1 ______________________________________ Dropout Information
Image Used ______________________________________ DCT F, B mode C,
B or F MVF C, B MVB C, F ______________________________________
When, as shown in Table 1 which is the error correspondence table,
a plural number of images are used when a frame of information has
no errors, and an error portion is discarded to use only a correct
portion when any frame of information has errors. When there are
errors in all of the frames which would have been used, there is
also the use of frames which would not have been used. In other
words, when an error is obtained in the motion vector MV of frame F
even though only the frame F is used in the prediction mode, the
intra-image prediction is performed by using the motion vector MV
of the frame B which was not originally used. Conversely, under the
condition where only the frame B is used in the prediction mode,
when the error is obtained in the frame B, the motion vector of the
frame F is used despite that the frame F was not originally used.
In the present invention, both motion vectors of the frames B and F
are transmitted together.
The operation of the variable adder 33 is the same as the residual
adder 20 for the conventional example, and when there is dropout of
the DCT information, the output of the intraframe decoder 21 is
made zero, and the output of the adaptive predictor 37 is output as
it is.
In this manner, according to this moving image coding apparatus and
decoding apparatus, the number of frames used for inter-image
processing is always made a plural number, and the information for
the motion compensation and inter-image processing method is
recombined into each type of information and transferred, with
errors for each type of information being detected and with
inter-image processing for each block having switching to another
frame without the use of signals of frames having errors in the
information, thereby lowering the probability that a plural number
of pieces of information in the same block will not be used even if
error detection is performed in coding units of a certain amount,
and thereby enabling there to be little deterioration of the image
quality.
The following is a description of a decoding apparatus of a second
embodiment.
The second embodiment of the present invention is applicable to the
apparatus "High-efficiency Coding Apparatus and Decoding
Apparatus," disclosed in U.S. Ser. No. 07/873,949 filed on Apr. 24,
1992, the inventor of which is the inventor of the invention of the
present application.
This embodiment has an improved coding efficiency while at the same
time maintaining frame independence, and therefore even uses the
inter-frame correlation for I-frames, and performs inter-frame
image addition at the decoding apparatus, so that the coding method
basically enables the image quality to be maintained even if the
quantization is rough.
The difference with the first embodiment is the decoding apparatus,
the configuration of which is shown in FIG. 5. The input data is
coded by the coding apparatus shown in FIG. 3. The difference
between the decoding apparatus of FIG. 5 and that of FIG. 4 is that
there is a matching decider 32 which decides the matching of two
images. More specifically, the configuration is such that the
output of the intraframe decoder 21 and the output of the
reproduced image signal output terminal 36 are both led to the
matching decider 32 and the variable adder 33.
In addition, the operation differs from the operation of the first
embodiment in that the P- and B-frames are the same for only the
I-frames. In the I-frames, the matching decider 32 checks the
matching of the two images, and gives that much information to the
variable adder 33. In the variable adder 33, the prediction signal
from the adaptive predictor 37 is increased when there is good
matching, while the current frame signal from the intraframe
decoder 21 is increased when there is poor matching, and adding is
then performed. Here, the sum of the gains of the respective
signals is "1".
When there is an error in the signal from the intraframe decoder,
the output of the intraframe decoder 21 is forcedly made "0" and
only the output of the adaptive predictor 37 is used as the
reproduced signals.
By this, it is possible to compensate for coding errors which have
occurred in independent frames.
Moreover, I-frame quantization error compensation can use a coding
apparatus in which the second embodiment of the present invention
has been applied.
* * * * *