U.S. patent application number 12/201521 was filed with the patent office on 2008-12-18 for contamination degree determining apparatus and contamination degree determining method.
This patent application is currently assigned to KABUSHIKI KAISHA TOSHIBA. Invention is credited to Naotake NATORI.
Application Number | 20080308244 12/201521 |
Document ID | / |
Family ID | 39721158 |
Filed Date | 2008-12-18 |
United States Patent
Application |
20080308244 |
Kind Code |
A1 |
NATORI; Naotake |
December 18, 2008 |
CONTAMINATION DEGREE DETERMINING APPARATUS AND CONTAMINATION DEGREE
DETERMINING METHOD
Abstract
A contamination degree determining apparatus for paper
determines a final contamination degree of paper on the basis of
values of various contamination characteristics indicating the
degrees of, for example, creases, wrinkles or stains obtained from
the paper. Values of a plurality of contamination characteristics
are input to the contamination degree determining apparatus for
paper. The apparatus calculates a plurality of evaluation values
from the plurality of input contamination characteristics, the
contribution degree of each of the plurality of contamination
characteristics, reference vectors reflecting the distribution of
the plurality of contamination characteristics, and dispersion
parameters reflecting the distribution of the plurality of
contamination characteristics. The apparatus determines the final
contamination degree of the paper on the basis of the calculated
evaluation values and the weight of each of the evaluation
values.
Inventors: |
NATORI; Naotake;
(Kunitachi-shi, JP) |
Correspondence
Address: |
PILLSBURY WINTHROP SHAW PITTMAN, LLP
P.O. BOX 10500
MCLEAN
VA
22102
US
|
Assignee: |
KABUSHIKI KAISHA TOSHIBA
Tokyo
JP
|
Family ID: |
39721158 |
Appl. No.: |
12/201521 |
Filed: |
August 29, 2008 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
PCT/JP2008/052994 |
Feb 21, 2008 |
|
|
|
12201521 |
|
|
|
|
Current U.S.
Class: |
162/198 ;
162/263 |
Current CPC
Class: |
G07D 7/187 20130101;
G06K 9/6273 20130101 |
Class at
Publication: |
162/198 ;
162/263 |
International
Class: |
D21F 7/00 20060101
D21F007/00 |
Foreign Application Data
Date |
Code |
Application Number |
Feb 22, 2007 |
JP |
2007-042292 |
Claims
1. A contamination degree determining apparatus comprising: an
input unit which inputs values of a plurality of contamination
characteristics detected from an inspection target; an evaluation
unit which calculates a plurality of evaluation values on the basis
of a contribution degree of each of the plurality of contamination
characteristics, reference vectors reflecting the distribution of
the plurality of contamination characteristics, and dispersion
parameters reflecting the distribution of the plurality of
contamination characteristics; and a determination unit which
determines a contamination degree of the inspection target on the
basis of the plurality of evaluation values calculated by the
evaluation unit.
2. The contamination degree determining apparatus according to
claim 1, further comprising: a storage unit which stores the
contribution degree of each of the plurality of contamination
characteristics.
3. The contamination degree determining apparatus according to
claim 1, further comprising: a setting unit which sets the
contribution degree of each of the plurality of contamination
characteristics on the basis of information input from an operation
device operated by an operator.
4. The contamination degree determining apparatus according to
claim 1, further comprising: an adjusting unit which
feedback-adjusts the contribution degree on the basis of the
determination result of the contamination degree by the
determination unit.
5. The contamination degree determining apparatus according to
claim 1, further comprising: a weight storage unit which stores a
weight corresponding to each of the evaluation values calculated by
the evaluation unit, wherein the determination unit determines the
contamination degree of the inspection target on the basis of the
plurality of evaluation values calculated by the evaluation unit
and the weight corresponding to each of the evaluation values
stored in the weight storage unit.
6. A contamination degree determining method comprising: inputting
values of a plurality of contamination characteristics detected
from an inspection target; calculating a plurality of evaluation
values on the basis of a contribution degree of each of the
plurality of contamination characteristics, reference vectors
reflecting the distribution of the plurality of contamination
characteristics, and dispersion parameters reflecting the
distribution of the plurality of contamination characteristics; and
determining a contamination degree of the inspection target on the
basis of the plurality of calculated evaluation values.
7. The contamination degree determining method according to claim
6, wherein the contribution degree of each of the plurality of
contamination characteristics is read from a storage unit.
8. The contamination degree determining method according to claim
6, wherein the contribution degree of each of the plurality of
contamination characteristics is set on the basis of information
input from an operation device operated by an operator.
9. The contamination degree determining method according to claim
6, further comprising: feedback-adjusting the contribution degree
on the basis of the determination result of the contamination
degree of the inspection target, wherein the contamination degree
of the inspection target is determined on the basis of the
plurality of evaluation values calculated by the feedback-adjusted
contribution degree of each of the plurality of contamination
characteristics, by the reference vectors and by the dispersion
parameters.
10. The contamination degree determining method according to claim
6, further comprising: reading a weight corresponding to each of
the evaluation values from a weight storage unit, wherein the
contamination degree of the inspection target is determined on the
basis of the plurality of calculated evaluation values and the
weight corresponding to each of the evaluation values.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This is a Continuation Application of PCT Application No.
PCT/JP2008/052994, filed Feb. 21, 2008, which was published under
PCT Article 21(2) in Japanese.
[0002] This application is based upon and claims the benefit of
priority from prior Japanese Patent Application No. 2007-042292,
filed Feb. 22, 2007, the entire contents of which are incorporated
herein by reference.
BACKGROUND OF THE INVENTION
[0003] 1. Field of the Invention
[0004] The present invention relates to a contamination degree
determining apparatus and a contamination degree determining method
for determining a final contamination degree of an inspection
target such as paper on the basis of a plurality of contamination
characteristics (such as creases, wrinkles or stains) detected from
the inspection target.
[0005] 2. Description of the Related Art
[0006] For example, Jpn. Pat. Appln. KOKAI Publication No. 7-190955
(Patent document 1) has disclosed a method of obtaining a value in
which the amounts of a plurality of contamination characteristics
(such as creases, wrinkles or stains) are weighted and linearly
coupled together, in determining a final contamination degree of
paper such as paper money from the contamination
characteristics.
[0007] However, the frequency distribution of contamination
characteristics is often nonlinear in general. In such a case, if
the contamination characteristics are simply linearly coupled
together as in the technique described in Patent document 1
mentioned above, the linearity of the frequency of the
contamination degree decreases, which may make it difficult to
obtain a desired percentage of contamination determination
(percentage of paper determined to be contaminated).
[0008] Furthermore, `D. E. Rumelhart, G. E. Hinton, R. J. Williams:
"Learning Representations by back-propagating errors", Nature 323,
pp. 533-536. 1986.` (Non Patent document 1) has disclosed a
technique which temporarily nonlinearly converts contamination
characteristics and linearly couples their values together to
improve the linearity of a contamination degree.
[0009] The technique described in Non Patent document 1 mentioned
above is a technique concerning the application of multilayer
perceptron type neural net and its learning method. In addition,
the technique of an artificial neural net has originally derived
from a simulation of a living body. The technique of the artificial
neural net is used in various industrial fields as a nonlinear
signal processing system from the aspect of engineering.
[0010] However, when the technique described in Non Patent document
1 mentioned above is applied to the determination of the
contamination degree of paper, results of learning the
contamination characteristics are represented by internal
parameters called weight vectors in a black box form (or in a
dispersed form). There is thus a problem that it is difficult to
adjust the contribution degrees of the respective contamination
characteristics in accordance with the form of operation.
BRIEF SUMMARY OF THE INVENTION
[0011] One aspect of this invention is directed to provide a
contamination degree determining apparatus and a contamination
degree determining method capable of accurately determining the
contamination degree of an inspection target in accordance with the
form of operation.
[0012] A contamination degree determining apparatus as one aspect
of this invention comprises: an input unit which inputs values of a
plurality of contamination characteristics detected from an
inspection target; an evaluation unit which calculates a plurality
of evaluation values on the basis of a contribution degree of each
of the plurality of contamination characteristics, reference
vectors reflecting the distribution of the plurality of
contamination characteristics, and dispersion parameters reflecting
the distribution of the plurality of contamination characteristics;
and a determination unit which determines a contamination degree of
the inspection target on the basis of the plurality of evaluation
values calculated by the evaluation unit.
[0013] A contamination degree determining method as one aspect of
this invention comprises: inputting values of a plurality of
contamination characteristics detected from an inspection target;
calculating a plurality of evaluation values on the basis of a
contribution degree of each of the plurality of contamination
characteristics, reference vectors reflecting the distribution of
the plurality of contamination characteristics, and dispersion
parameters reflecting the distribution of the plurality of
contamination characteristics; and determining a contamination
degree of the inspection target on the basis of the plurality of
calculated evaluation values.
BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWING
[0014] FIG. 1 is a diagram schematically showing an example of the
configuration of a paper processing system to which a contamination
degree determining apparatus for paper according to each embodiment
is applied;
[0015] FIG. 2 is a block diagram schematically showing an example
of the configuration of a contamination degree determining
apparatus for paper according to a first embodiment;
[0016] FIG. 3 is a flowchart for explaining the flow of the whole
processing in the contamination degree determining apparatus
according to the first embodiment;
[0017] FIG. 4 is a flowchart for explaining input processing for
inputting contamination characteristics;
[0018] FIG. 5 is a flowchart for explaining contribution degree
read processing;
[0019] FIG. 6 is a flowchart for explaining reference vector read
processing;
[0020] FIG. 7 is a flowchart for explaining dispersion parameter
read processing;
[0021] FIG. 8 is a flowchart for explaining evaluation value
calculating processing;
[0022] FIG. 9 is a flowchart for explaining weight read
processing;
[0023] FIG. 10 is a flowchart for explaining contamination degree
determining processing for determining a final contamination degree
of one sheet of paper;
[0024] FIG. 11 is a schematic diagram for explaining a local
representation type neural net;
[0025] FIG. 12 is a diagram showing a reaction characteristic of an
intermediate layer unit of the local representation type neural
net;
[0026] FIG. 13 is a block diagram schematically showing an example
of the configuration of a contamination degree determining
apparatus for paper according to a second embodiment;
[0027] FIG. 14 is a flowchart for explaining the flow of the whole
processing in the contamination degree determining apparatus
according to the second embodiment;
[0028] FIG. 15 is a flowchart for explaining contribution degree
setting processing;
[0029] FIG. 16 is a block diagram schematically showing an example
of the configuration of a contamination degree determining
apparatus for paper according to a third embodiment;
[0030] FIG. 17 is a flowchart for explaining the flow of the whole
processing in the contamination degree determining apparatus
according to the third embodiment; and
[0031] FIG. 18 is a flowchart for explaining feedback adjusting
processing for contribution degrees.
DETAILED DESCRIPTION OF THE INVENTION
[0032] Hereinafter, embodiments of the present invention will be
described with reference to the drawings.
[0033] First explained is a paper processing system to which a
contamination degree determining apparatus as each embodiment
described below is applied.
[0034] FIG. 1 is a diagram schematically showing an example of the
configuration of the paper processing system.
[0035] The paper processing system shown in FIG. 1 processes paper
as an inspection target on the basis of the result of determining a
contamination degree. The paper processing system shown in FIG. 1
comprises a contamination degree determining apparatus 1 (1A, 1B,
1C), a conveyer 2, a conveyance controller 3, various sensors S
(S1, S2, . . . Sn), a monitor 4 and an operation device 5.
[0036] The conveyer 2 is a device for conveying paper P to be an
inspection target (target for contamination degree determination).
The conveyance controller 3 controls the conveyance of the paper P
by the conveyer 2. Each of the sensors S1, S2, . . . Sn is a sensor
for detecting various contamination characteristics from the paper
P conveyed by the conveyer 2.
[0037] The contamination degree determining apparatus 1 determines
the contamination degree of the paper P as the inspection target.
The contamination degree determining apparatus 1 determines the
contamination degree of the paper P on the basis of values (various
contamination characteristics) detected from the paper P by the
sensors. The contamination degree determining apparatus 1 is
configured by, for example, a computer capable of data processing
by the execution of an application program.
[0038] For example, the contamination degree determining apparatus
1 has a control unit 6, a storage unit 7 and various interfaces
(not shown), as shown in FIG. 1. In the contamination degree
determining apparatus 1, the control unit 6 executes a program
stored in the storage unit 7 to achieve various kinds of processing
(functions) described later. For example, the control unit 6
executes the program stored in the storage unit 7 such that
contamination degree determining processing for paper described in
each embodiment is achieved.
[0039] Furthermore, the contamination degree determining apparatus
1 is connected to the conveyance controller 3, the monitor 4 and
the operation device 4 via various unshown interfaces or to the
sensors S. Thus, the contamination degree determining apparatus 1
controls the conveyance state of the paper P conveyed by the
conveyer 2 via the conveyance controller 3, displays information on
the monitor 4 or acquires information input by an operator through
the operation device 5.
[0040] First, second and third embodiments of the contamination
degree determining apparatuses 1A, 1B, 1C applied to the
above-mentioned paper processing system are described below.
[0041] The first embodiment is initially described.
[0042] FIG. 2 schematically shows an example of the configuration
of the contamination degree determining apparatus 1A for paper
(inspection target) according to the first embodiment.
[0043] The contamination degree determining apparatus 1A for paper
shown in FIG. 2 has an input unit 11, a contribution degree storage
unit 12, a reference vector storage unit 13, a dispersion parameter
storage unit 14, an evaluation unit 15, a weight storage unit 16
and a determination unit 17.
[0044] Here, the contribution degree storage unit 12, the reference
vector storage unit 13, the dispersion parameter storage unit 14
and the weight storage unit 16 are realized as storage areas in the
storage unit 7. The evaluation unit 15 and the determination unit
17 are functions achieved when the control unit 6 executes the
program stored in the storage unit 7.
[0045] The input unit 11 inputs values of a plurality of
contamination characteristics (e.g., values indicating the degrees
of creases, wrinkles, stains) obtained from paper as such paper
money. For example, the input unit 11 inputs information detected
by the sensors S. Moreover, an unshown external device may generate
values indicating contamination characteristics from the
information detected by the sensors S, and the generated values may
be input to the input unit 11.
[0046] The contribution degree storage unit 12 stores the
contribution degree of each of the plurality of contamination
characteristics. The reference vector storage unit 13 stores
reference vectors reflecting the distribution of the plurality of
contamination characteristics. The dispersion parameter storage
unit 14 stores dispersion parameters reflecting the distribution of
the plurality of contamination characteristics. The contribution
degrees stored in the contribution degree storage unit 12, the
reference vectors stored in the reference vector storage unit 13,
and the dispersion parameters stored in the dispersion parameter
storage unit 14 are described later in detail.
[0047] The evaluation unit 15 calculates an evaluation value from
the contamination characteristics input by the input unit 11, the
contribution degrees stored in the contribution degree storage unit
12, the reference vectors stored in the reference vector storage
unit 13, and the dispersion parameters stored in the dispersion
parameter storage unit 14. Further, the evaluation unit 15
calculates an evaluation value for each of the various
contamination characteristics input by the input unit 11. The
weight storage unit 16 stores a weight for each evaluation value.
The determination unit 17 determines a final contamination degree
of the paper on the basis of evaluation value calculated by the
evaluation unit 15 and the weights stored in the weight storage
unit 16.
[0048] Next, processing executed by the contamination degree
determining apparatus for paper as the first embodiment having the
above-mentioned configuration is described.
[0049] FIG. 3 is a flowchart for explaining the flow of the whole
processing in the contamination degree determining apparatus 1A for
paper.
[0050] First, the input unit 11 performs processing for inputting I
contamination characteristics x.sub.i to the evaluation unit 15
(step S1). Here, the contamination characteristics mean values
(characteristic amounts) quantitatively representing the various
degrees of contaminations in paper. For example, the contamination
characteristics are values (amounts of sensation) in which, for
example, creases, wrinkles or stains are converted into numerical
values, or measurement amounts such as density values or
differential values.
[0051] FIG. 4 is a flowchart for explaining input processing for
inputting I contamination characteristics. Here, the contamination
characteristic input processing is explained with reference to FIG.
4.
[0052] As shown in FIG. 4, first, "1" is set to the input unit 11
as an initial value of a variable i (step S11). After setting the
variable i at "1", the input unit 11 checks whether or not the
variable i is equal to or less than a predetermined value I (step
S12). If the variable i is not equal to or less than I, that is, if
the variable i is greater than I (step S12, NO), the input unit 11
terminates the input processing of the contamination characteristic
x.sub.i.
[0053] If the variable i is not equal to or less than I (step S12,
YES), the input unit 11 inputs the contamination characteristic
x.sub.i (step S13). After the contamination characteristic x.sub.i
is input, the input unit 11 increments the variable i ("i=i+1")
(step S14), and returns to step S12. That is, the input unit 11
repeatedly executes the processing in steps S12 to S14 until the
variable i is greater than I. Thus, the input unit 11 inputs I
contamination characteristics, and finishes the input
processing.
[0054] When the input processing by the input unit 11 is finished,
the evaluation unit 15 performs read processing for reading a
contribution degree m.sub.i for each of I contamination
characteristics x.sub.i from the contribution degree storage unit
12 (step S2). Here, a contribution degree is set for each of the
contamination characteristics. In other words, the degree of
contribution of a certain contamination characteristic to the
determination of a final contamination degree is only determined by
this value.
[0055] FIG. 5 is a flowchart for explaining processing for reading
I contribution degrees (contribution degree read processing). Here,
the contribution degree read processing is explained with reference
to FIG. 5. That is, as the contribution degree read processing, the
evaluation unit 15 first sets "1" as the initial value of the
variable i (step S21). After setting the variable i at "1", the
evaluation unit 15 checks whether or not the variable i is equal to
or less than I. If the variable i is not equal to or less than I,
that is, if the variable i is greater than I (step S22, NO), the
evaluation unit 15 terminates the contribution degree read
processing.
[0056] If the variable i is equal to or less than I (step S22,
YES), the evaluation unit 15 reads the contribution degree m.sub.i
(step S23). After reading the contribution degree m.sub.i, the
evaluation unit 15 increments the variable i ("i=i+1") (step S24),
and returns to step S22. That is, the evaluation unit 15 repeatedly
executes the processing in steps S22 to S24 until the variable i is
greater than I. Thus, the evaluation unit 15 reads I contribution
degrees, and finishes the contribution degree read processing.
[0057] Then, the evaluation unit 15 reads
p.sub.j [Equation 1]
[0058] J I-dimensional reference vectors from the reference vector
storage unit 13 (step S3). Here, the reference vectors and the
dispersion parameters described later reflect the distributions of
the contamination characteristics in an I-dimensional space formed
by I contamination characteristics. That is, the distribution of
the contamination characteristics can be modeled by a mixed
distribution, and probability density distributions constituting
the distribution are represented by the above-motioned reference
vectors or the later-described dispersion parameters.
[0059] FIG. 6 is a flowchart for explaining processing for reading
J I-dimensional reference vectors (reference vector read
processing). Here, the reference vector read processing is
explained with reference to FIG. 6.
[0060] That is, as the reference vector read processing, the
evaluation unit 15 first sets "1" as an initial value of a variable
j (step S31). After setting the variable j at "1", the evaluation
unit 15 checks whether or not the variable j is equal to or less
than the J (step S32).
[0061] If the variable j is equal to or less than J (step S32,
YES), the evaluation unit 15 sets "1" as an initial value of a
variable i (step S33). After setting the variable i at "1", the
evaluation unit 15 checks whether or not the variable i is equal to
or less than I (step S34).
[0062] If the variable i is equal to or less than I (step S34,
YES), the evaluation unit 15 reads a reference vector p.sub.ji
(step S35). After reading the reference vector p.sub.ji, the
evaluation unit 15 increments the variable i ("i=i+1") (step S36),
and returns to step S34.
[0063] If the variable i is not equal to or less than I, that is,
if the variable i is greater than I (step S34, NO), the evaluation
unit 15 increments the variable j ("j=j+1") (step S37), and returns
to step S32. That is, the evaluation unit 15 repeatedly executes
the processing in steps S34 to S36 until the variable i is greater
than I. By the processing in steps S34 to S36, the evaluation unit
15 reads I reference vectors for the particular variable j.
[0064] Furthermore, if the variable j is not equal to or less than
J, that is, if the variable j is greater than J (step S32, NO), the
evaluation unit 15 terminates the reference vector read processing.
That is, the evaluation unit 15 repeatedly executes the processing
in steps S32 to S37 until the variable j is greater than J. By the
processing in steps S32 to S37, the evaluation unit 15 reads I
reference vectors for all the values (1 to J) of the variable j. As
a result, the evaluation unit 15 reads J.times.I reference vectors
(J I-dimensional reference vectors).
[0065] Then, the evaluation unit 15 reads J dispersion parameters
s.sub.j from the dispersion parameter storage unit 14 (step
S4).
[0066] FIG. 7 is a flowchart for explaining processing for reading
J dispersion parameters (dispersion parameter read processing).
Here, the dispersion parameter read processing is explained with
reference to FIG. 7.
[0067] That is, as the dispersion parameter read processing, the
evaluation unit 15 first sets "1" as an initial value of a variable
j (step S41). After setting the variable j at "1", the evaluation
unit 15 checks whether or not the variable j is equal to or less
than a predetermined value J. If the variable j is not equal to or
less than J, that is, if the variable j is greater than J (step
S42, NO), the evaluation unit 15 terminates the dispersion
parameter read processing.
[0068] If the variable j is not equal to or less than J (step S42,
YES), the evaluation unit 15 reads a dispersion parameter s.sub.j
(step S43). After reading the dispersion parameter s.sub.j, the
evaluation unit 15 increments the variable j ("j=j+1") (step S44),
and returns to step S42. That is, the evaluation unit 15 repeatedly
executes the processing in steps S42 to S44 until the variable j is
greater than J. Thus, the evaluation unit 15 reads J dispersion
parameters, and finishes the dispersion parameter read
processing.
[0069] Then, the evaluation unit 15 performs evaluation value
calculating processing for calculating J evaluation values y.sub.j
(step S5). J evaluation values y.sub.j are calculated on the basis
of the contamination characteristics x.sub.i, the contribution
degrees m.sub.i, the reference vectors
p.sub.j [Equation 2]
and the dispersion parameters s.sub.j. Calculations of J evaluation
values y.sub.j are sent from the evaluation unit 15 to the
determination unit 17.
[0070] An equation for calculating the evaluation value is, for
example, as follow:
y j = exp ( - u j 2 s j 2 ) , u j = i = 1 I m i ( x i - p ji ) 2 [
Equation 3 ] ##EQU00001##
wherein u.sub.j is the weighted square sum of the difference
between the contamination characteristic x.sub.i and the reference
vector element p.sub.ji due to the contribution degree m.sub.i.
[0071] FIG. 8 is a flowchart for explaining evaluation value
calculating processing for calculating J evaluation values y.sub.j.
Here, the evaluation value calculating processing is explained with
reference to FIG. 8.
[0072] That is, as the evaluation value calculating processing, the
evaluation unit 15 first sets "1" as an initial value of a variable
j (step S51). After setting the variable j at "1", the evaluation
unit 15 checks whether or not the variable j is equal to or less
than a predetermined value J (step S52).
[0073] If the variable j is equal to or less than J (step S52,
YES), the evaluation unit 15 sets the weighted square sum u.sub.j
to "0" as an initial value (step S53), and sets "1" as an initial
value of a variable i (step S54). After setting "1" as the initial
value of the variable i, the evaluation unit 15 checks whether or
not the variable i is equal to or less than I (step S55).
[0074] If the variable i is equal to or less than I (step S55,
YES), the evaluation unit 15 calculates the weighted square sum
u.sub.j for the variable i in accordance with a calculation
equation "u.sub.j=u.sub.j+m.sub.i(x.sub.i-p.sub.ji).sup.2" (step
S56). Thus, after calculating the weighted square sum u.sub.j for
the variable i, the evaluation unit 15 increments the variable i
("i=i+1") (step S57), and returns to step S55.
[0075] Furthermore, if the variable i is not equal to or less than
I, that is, if the variable i is greater than I (step S55, NO), the
evaluation unit 15 calculates the evaluation value y.sub.j for the
variable j in accordance with a calculation equation
"y.sub.j=exp(-U.sub.j/2s.sub.j.sup.2)" (step S58). After
calculating the evaluation value y.sub.j for the variable j, the
evaluation unit 15 increments the variable j ("j=j+1") (step S59),
and returns to step S52.
[0076] That is, the evaluation unit 15 repeats the processing in
steps S55 to S57 until the variable i is greater than I, and
calculates the evaluation value y.sub.j for the variable j in
accordance with step S58. That is, by the processing in steps S55
to S58, the evaluation unit 15 calculates the evaluation value
y.sub.j for the variable j.
[0077] Furthermore, if the variable j is not equal to or less than
J, that is, if the variable j is greater than J (step S52, NO), the
evaluation unit 15 terminates the evaluation value calculating
processing. That is, the evaluation unit 15 repeatedly executes the
processing in steps S52 to S59 until the variable j is greater than
J. Thus, the evaluation unit 15 calculates the evaluation values
y.sub.j for all the values (1 to J) of the variable j.
[0078] Then, the determination unit 17 reads J weights w.sub.j from
the weight storage unit 16 (step S6). Here, a weight is set for
each of J evaluation values. That is, the weight corresponds to a
"mixed parameter" in a mixed distribution model.
[0079] FIG. 9 is a flowchart for explaining processing for reading
J weights w.sub.j (weight read processing). Here, the weight read
processing is explained with reference to FIG. 9.
[0080] That is, as the weight read processing, the determination
unit 17 first sets "1" as an initial value of a variable j (step
S61). After setting the variable j at "1", the determination unit
17 checks whether or not the variable j is equal to or less than J
(step S62). If the variable j is not equal to or less than J, that
is, if the variable j is greater than J (step S62, NO), the
determination unit 17 terminates the weight read processing.
[0081] Furthermore, if the variable j is equal to or less than J
(step S62, YES), the determination unit 17 reads the weight w.sub.j
(step S63). After reading the weight w.sub.j, the determination
unit 17 increments the variable j ("j=j+1") (step S64), and returns
to step S62. That is, the determination unit 17 repeatedly executes
the processing in steps S62 to S64 until the variable j is greater
than J. Thus, the determination unit 17 reads J weights, and
finishes the weight read processing.
[0082] Then, the determination unit 17 determines a final
contamination degree z of one sheet of paper targeted for the
determination of the contamination degree (step S7). The
contamination degree z is determined on the basis of the evaluation
value y.sub.j calculated by the evaluation unit 15 and the weight
w.sub.j read from the weight storage unit 16. In addition, this
contamination degree z is a weighted linear coupling of J
evaluation values (not nonlinearly converted)
[0083] The final contamination degree z is calculated by, for
example, the following calculation equation:
z = j = 1 J w j y j [ Equation 4 ] ##EQU00002##
[0084] FIG. 10 is a flowchart for explaining processing for
determining a final contamination degree of one sheet of paper.
Here, the contamination degree determining processing is explained
with reference to FIG. 10.
[0085] That is, as the contamination degree determining processing,
the determination unit 17 first sets "1" as an initial value of a
variable j (step S71). After setting the variable j at "1", the
determination unit 17 sets "z=0" as an initial value of the
contamination degree z (step S72). After setting the contamination
degree z, the determination unit 17 checks whether or not the
variable j is equal to or less than J (step S73).
[0086] Furthermore, if the variable j is equal to or less than J
(step S73, YES), the determination unit 17 calculates the
contamination degree z in accordance with a calculation equation
"z=z+w.sub.j*y.sub.j" (step S74). After calculating the
contamination degree z, the determination unit 17 increments the
variable j ("j=j+1") (step S75), and returns to step S73. That is,
the determination unit 17 repeatedly executes the processing in
steps S73 to S75 until the variable j is greater than J.
[0087] Furthermore, if the variable j is not equal to or less than
J, that is, if the variable j is greater than J (step S73, NO), the
determination unit 17 determines the contamination degree z in the
case of j=J to be the final contamination degree z of the paper
(step S76).
[0088] In accordance with the processing described above, the
contamination degree determining apparatus 1A according to the
first embodiment introduces a static contribution degree to each of
the various contamination characteristics in order to determine the
contamination degree of each sheet of paper. This makes it possible
for the contamination degree determining apparatus 1A to readily
achieve optimum weighting of each of the contamination
characteristics for determining a comprehensive contamination
degree of paper in accordance with the form of operation.
[0089] In addition, the first embodiment described above and the
second and third embodiments described later are based on the idea
of a local representation type neural net. The local representation
type neural net is introduced in, for example, Non Patent document
2 below.
[0090] `J. E. Moody and C. Darken: "Fast learning in networks of
locally-tuned processing units", Neural Computation 1, pp. 281-294.
1989.` (Non Patent document 2).
[0091] The concept of the local representation type neural net is
described below.
[0092] FIG. 11 is a schematic diagram for explaining the local
representation type neural net. Further, FIG. 12 is a diagram
showing a reaction characteristic of an intermediate layer unit of
the local representation type neural net.
[0093] Three layers shown in FIG. 11 are an input layer, an
intermediate layer (hidden layer) and an output layer. There are
one or more units in each layer. Here, the number of input layer
units is a number (I) corresponding to the contamination
characteristics, and the number of output layer units is one
corresponding to the final contamination degree. In addition, the
number of intermediate layer units is a number (J) sufficient to
obtain desired performance.
[0094] In the local representation type neural net, there are no
intra-layer couplings and there are inter-layer couplings alone,
and a signal is transmitted from the input layer to the output
layer in a feedforward manner, as in the three-layer perceptron
type neural net described in Non Patent document 1. A reaction
characteristic xi of the input layer unit is exactly the value
(identity function) of the input contamination characteristics. A
reaction characteristic yj of the intermediate layer unit is not a
characteristic like a sigmoid function used in the three-layer
perceptron type neural net, and is a probability density
distribution type characteristic as shown in FIG. 12. A reaction
characteristic z of the output layer unit is a linear coupling.
[0095] Next, the second embodiment will be described.
[0096] FIG. 13 is a diagram schematically showing an example of the
configuration of the contamination degree determining apparatus 1B
for paper according to the second embodiment.
[0097] The contamination degree determining apparatus 1B shown in
FIG. 13 has a configuration in which the contribution degree
storage unit 12 in the contamination degree determining apparatus
1A shown in FIG. 2 is replaced with a contribution degree setting
unit 18 for setting a contribution degree of each of a plurality of
contamination characteristics. That is, the configuration of the
contamination degree determining apparatus 1B is the same as that
of the contamination degree determining apparatus 1A described in
the first embodiment except for the contribution degree setting
unit 18. Therefore, like signs are assigned to like parts, and the
configuration of the contamination degree determining apparatus 1B
similar to that of the contamination degree determining apparatus
1A is not described in detail.
[0098] The contribution degree setting unit 18 sets, for an
evaluation unit 15, a contribution degree of each of a plurality of
contamination characteristics input by the operator using the
operation device 5. In addition, the contribution degree setting
unit 17 can be realized, for example, when the control unit 6
executes the program stored in the storage unit 7 in the hardware
configuration shown in FIG. 1.
[0099] That is, the contamination degree determining apparatus 1B
does not accumulate the contribution degrees of the contamination
characteristics in contrast with the contamination degree
determining apparatus 1A described in the first embodiment, and the
contribution degree setting unit 18 sets a contribution degree of
each of a plurality of contamination characteristics in accordance
with the setting operation by the operator. In other words, the
operator is able to set (adjust) the contamination degree in the
contamination degree determining apparatus 1B.
[0100] FIG. 14 is a flowchart showing the flow of the whole
processing in the contamination degree determining apparatus 1B
according to the second embodiment. In the flowchart shown in FIG.
14, step S2 for reading a contribution degree in the flowchart
shown in FIG. 3 is replaced with step S8 for contamination degree
setting (adjusting) processing. Therefore, in the flowchart shown
in FIG. 14, the same signs are assigned to the same parts as those
in the flowchart shown in FIG. 3, and these parts are not described
in detail.
[0101] Here, the contamination degree setting processing in step S8
is processing to set (adjust) I contamination degrees input by the
operator through, for example, the operation device 5. FIG. 15 is a
flowchart for explaining contribution degree setting processing for
setting (adjusting) I contribution degrees. The contribution degree
setting processing is explained below with reference to FIG.
15.
[0102] That is, as the contribution degree setting processing, the
contribution degree setting unit 18 first sets "1" as an initial
value of a variable i (step S81). After setting the variable i at
"1", the contribution degree setting unit 18 checks whether or not
the variable i is equal to or less than I (step S82).
[0103] If the variable i is equal to or less than I (step S82,
YES), the contribution degree setting unit 18 sets (adjusts) a
contribution degree m.sub.i in accordance with the input operation
to the operation device 5 by the operator (step S83). After setting
the contribution degree m.sub.i, the contribution degree setting
unit 18 increments the variable i ("i=i+1") (step S84), and returns
to step S82. If the variable i is not equal to or less than I, that
is, if the variable i is greater than I (step S82, NO), the
contribution degree setting unit 18 terminates the contribution
degree setting processing.
[0104] That is, the contribution degree setting unit 18 repeatedly
executes the processing in steps S82 to S84 until the variable i is
greater than I. Thus, the contribution degree setting unit 18
finishes the processing for setting I contribution degrees.
[0105] When I contribution degrees are set by the contribution
degree setting processing described above, the contamination degree
determining apparatus 1B determines a final contamination degree of
paper P targeted for determination, in accordance with the flow of
the processing described in the first embodiment.
[0106] Furthermore, in the contamination degree determining
apparatus 1B according to the second embodiment, the contribution
degree m.sub.i is independent of a reference vector representing a
model
p.sub.j [Equation 5]
or a dispersion parameter s.sub.j. That is, a contribution degree
is statically set for each dimension of the contamination
characteristics in the contamination degree determining apparatus
1B. Therefore, the value of the contribution degree m.sub.i does
not change (e.g., fixed at, for example, .A-inverted.m.sub.i=1) due
to the acquisition (learning) of a model, and is set or adjusted by
the operator as needed in accordance with the type of operation
after learning (during operation).
[0107] In addition, the second embodiment may be combined with the
first embodiment described above to determine the contamination
degrees. For example, as to the contribution degrees of the
contamination characteristics, the contribution degrees of some of
the contamination characteristics alone may be set to the
contribution degrees input by the operator, and the contribution
degrees of the other contamination characteristics may be read from
the storage unit.
[0108] Next, the third embodiment will be described.
[0109] FIG. 16 is a diagram schematically showing an example of the
configuration of the contamination degree determining apparatus 1C
for paper according to the third embodiment.
[0110] In the configuration of the contamination degree determining
apparatus 1C shown in FIG. 16, a feedback adjusting unit 19 for
feedback-adjusting an initially set contribution degree in
accordance with the determination result in a determination unit 17
is added to the contamination degree determining apparatus 1A shown
in FIG. 2 described in the first embodiment. That is, the
configuration of the contamination degree determining apparatus 1C
is the same as that of the contamination degree determining
apparatus 1A described in the first embodiment except for the
feedback adjusting unit 19. Therefore, like signs are assigned to
like parts, and the configuration of the contamination degree
determining apparatus 1B similar to that of the contamination
degree determining apparatus 1A is not described in detail.
[0111] The feedback adjusting unit 19 adjusts the contribution
degree in accordance with the determination result of the
contamination degree. That is, the feedback adjusting unit 19
feedback-adjusts an initially set (e.g., at .A-inverted.m.sub.i=1)
contribution degree of each of the contamination characteristics in
accordance with the determination result of the contamination
degree. For example, the feedback adjusting unit 19 calculates a
degree of influence of each of the contamination characteristics,
and make an adjustment so that the contribution degrees of the
contamination characteristics with lower degrees of influence may
be lower. Moreover, one way to calculate an influence degree
E.sub.i of the contamination characteristic x.sub.i is, for
example, to use an output difference when the contribution degree
of the contamination characteristic is temporarily set at zero as
in the following equation.
E.sub.i=z-z.sub.mi=0
[0112] In addition, the feedback adjusting unit 19 can be realized,
for example, when the control unit 6 executes the program stored in
the storage unit 7 in the hardware configuration shown in FIG.
1.
[0113] Next, the processing in the contamination degree determining
apparatus 1C is described.
[0114] FIG. 17 is a flowchart showing the flow of the whole
processing in the contamination degree determining apparatus 1C
according to the third embodiment.
[0115] That is, the evaluation unit 15 first reads the contribution
degree m.sub.i of each of I contamination characteristics from the
contribution degree storage unit 12, and sets the read contribution
degree m.sub.i as an initial value (step S91). The reading of the
contribution degree m.sub.i is similar to the above-mentioned
processing in step S2 in FIG. 3. That is, the contamination degree
determining apparatus 1C according to the third embodiment starts
the processing using, as an initial value, the contribution degree
m.sub.i of each of the contamination characteristics stored in the
contribution degree storage unit 12.
[0116] Then, the input unit 11 inputs I contamination
characteristics X.sub.i to the evaluation unit 15 (step S92). The
input of the contamination characteristics X.sub.i is similar to
the above-mentioned processing in step S1 in FIG. 3. Subsequently,
the contamination degree determining apparatus 1C sequentially
carries out the reference vector read processing (step S93), the
dispersion parameter read processing (step S94), the evaluation
value calculating processing (step S95), the weight read processing
(step S96) and the contamination degree determining processing
(step S97). The processing in steps S93 to S97 is similar to the
above-mentioned processing in steps S3 to S7 in FIG. 3.
[0117] That is, when the contamination degree z of paper targeted
for determination is determined by the determination unit 17, the
feedback adjusting unit 19 performs feedback adjusting processing
for feedback-adjusting the contribution degree of each of the
contamination characteristics set in the evaluation unit 15 in
accordance with the determination result in the determination unit
17 (step S98). In this feedback adjusting processing, as described
above, for example, a degree of influence of each of the
contamination characteristics is calculated, and an adjustment is
made so that the contribution degrees of the contamination
characteristics with lower degrees of influence may be lower.
[0118] After calculating an adjustment value of the contamination
degree, the feedback adjusting unit 19 checks whether or not a
feedback adjustment terminating condition (a preset number of
feedback adjustments) has been reached (step S99). When the
feedback adjustment terminating condition has not been reached, the
contamination degree determining apparatus 1C returns to step S92,
and repeatedly executes the processing in steps S92 to S99. When
the feedback adjustment terminating condition has been reached, the
contamination degree determining apparatus 1C terminates the
processing.
[0119] Next, the feedback adjusting processing is described in
detail.
[0120] FIG. 18 is a flowchart for explaining the feedback adjusting
processing for feedback-adjusting I contribution degrees. The
feedback adjusting processing is explained below with reference to
FIG. 18.
[0121] As shown in FIG. 18, in the feedback adjusting processing,
the feedback adjusting unit 19 first sets "1" as an initial value
to a variable i (step S101). After setting the variable i, the
feedback adjusting unit 19 checks whether or not the variable i is
equal to or less than I (step S102).
[0122] If the variable i is equal to or less than I (step S102,
YES), the feedback adjusting unit 19 checks whether or not the
influence degree of the contamination characteristic x.sub.i is low
(step S103).
[0123] When the influence degree of the contamination
characteristic x.sub.i is determined to be low as a result of the
check, the feedback adjusting unit 19 performs processing for
decreasing the value of the contribution degree m.sub.i. For
example, the feedback adjusting unit 19 performs processing for
decreasing the value of the contribution degree m.sub.i by
.DELTA.m.sub.i (processing for achieving
[m.sub.i=m.sub.i-.DELTA.m.sub.i]) (step S104). After performing the
processing for decreasing the contribution degree m.sub.i, the
feedback adjusting unit 19 increments the variable i ("i=i+1")
(step S105), and returns to step S102.
[0124] Furthermore, when the influence degree of the contamination
characteristic x.sub.i is determined not to be low as a result of
the check in step S103, the feedback adjusting unit 19 jumps the
processing for decreasing the contribution degree m.sub.i (i.e.,
the processing in step S104), and increments the variable i
("i=i+1") (step S105), and then returns to step S102.
[0125] That is, the feedback adjusting unit 19 repeatedly executes
the processing in steps S102 to S105 until the variable i is
greater than I.
[0126] If the variable i is determined to be not equal to or less
than I, that is, if the variable i is determined to be greater than
I (step S102, NO), the feedback adjusting unit 19 re-normalizes all
the contribution degrees m.sub.i (step S106), and updates all the
contribution degrees m.sub.i set in the evaluation unit 15 in
accordance with all the normalized contribution degrees
m.sub.i.
[0127] As described above, in the contamination degree determining
apparatus 1C according to the third embodiment, each contribution
degree of each of the contamination characteristics is optimized in
accordance with the determination result of the contamination
degree. Moreover, as described above, the results of the feedback
adjustment are easily understandable to the operator.
[0128] In addition, the third embodiment is not limited to this,
and can be carried out in various forms. For example, in the
feedback adjustment of the contribution degrees, the contribution
degrees of the contamination characteristics with higher degrees of
influence may be higher, or the contribution degrees can be
optimized by values corresponding to the influence degrees.
Moreover, the third embodiment described above may be carried out
in combination with the second embodiment. In this case, it is
possible to use the contribution degree set by the operator in step
S91 as an initial value to achieve the above-mentioned processing
shown in FIG. 17.
[0129] As described above, according to the embodiments described
above, the degree of contribution to the determination of a final
contamination degree is independently defined for each of the input
contamination characteristics. It is thus possible to provide a
contamination degree determining apparatus capable of adapting to
the weighting of the contamination characteristics corresponding to
the form of operation.
[0130] Furthermore, in the contamination degree determining
apparatus according to the second embodiment, the value of the
contribution degree of each of the contamination characteristics
does not change during learning. It is thus possible to set or
adjust the value of the contribution degree as needed in accordance
with the type of operation after learning (during operation).
[0131] Still further, in the contamination degree determining
apparatus according to the third embodiment, the contribution
degree of each of the contamination characteristics is
automatically optimized in accordance with the determination result
of the contamination degree. Thus, for example, when a client does
not know his own demand, a potential demand of the client can be
extracted simply by evaluating a sample, which is easily
understandable to the client.
[0132] It is to be noted that the contamination degree determining
apparatus and the contamination degree determining method according
to each of the embodiments are not limited to the determination of
the contamination degree of paper, and the apparatus can be applied
to the determination of the contamination degree of a target other
than paper. Moreover, the contamination degree determining
apparatus and the contamination degree determining method according
to each of the embodiments are not limited to the determination of
the contamination degree. That is, the determining apparatus and
the determining method according to each of the embodiments can be
applied to detect various characteristic amounts from an inspection
target and comprehensively evaluate the characteristics of the
inspection target on the basis of the detected various
characteristic amounts.
[0133] According to one aspect of the present invention, it is
possible to provide a contamination degree determining apparatus
and a contamination degree determining method capable of accurately
determining the contamination degree of an inspection target in
accordance with the form of operation.
* * * * *