CN107122795A - A kind of pedestrian integrated based on coring feature and stochastic subspace discrimination method again - Google Patents
A kind of pedestrian integrated based on coring feature and stochastic subspace discrimination method again Download PDFInfo
- Publication number
- CN107122795A CN107122795A CN201710212251.XA CN201710212251A CN107122795A CN 107122795 A CN107122795 A CN 107122795A CN 201710212251 A CN201710212251 A CN 201710212251A CN 107122795 A CN107122795 A CN 107122795A
- Authority
- CN
- China
- Prior art keywords
- msub
- mrow
- mover
- pedestrian
- coring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000012850 discrimination method Methods 0.000 title claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims abstract description 26
- 238000012360 testing method Methods 0.000 claims abstract description 25
- 238000012549 training Methods 0.000 claims abstract description 25
- 238000005315 distribution function Methods 0.000 claims abstract description 10
- 238000000034 method Methods 0.000 claims description 25
- 230000008569 process Effects 0.000 claims description 5
- 239000012141 concentrate Substances 0.000 claims description 4
- 230000006870 function Effects 0.000 description 11
- 230000000007 visual effect Effects 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 4
- 241000271897 Viperidae Species 0.000 description 3
- 238000013480 data collection Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 201000007094 prostatitis Diseases 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of pedestrian integrated based on coring feature and stochastic subspace discrimination method again, comprise the following steps：S1, obtains the training sample set and test sample collection of pedestrian image, determines the coring function between two samples；S2, is converted into coring feature by the primitive character of two sample sets respectively；S3, in the coring feature space of training sample set, randomly selects multiple different subspaces, and the covariance matrix and its inverse matrix of the coring feature difference of different and identical pedestrian image pair are calculated respectively, the distribution function of the coring feature difference of image pair is obtained；S4, respectively under each subspace, calculate sample to the probability for identical pedestrian and be different pedestrians probability, regard the ratio of two probability as the distance between sample；S5, progress of adjusting the distance is integrated, obtains the final distance between each sample pair.Compared with prior art, the present invention has good pedestrian identification capability again, it is adaptable to a variety of features, with stronger robustness.
Description
Technical field
The present invention relates to the feature extraction in monitor video intellectual analysis and learning distance metric method, more particularly, to one
Kind based on the integrated pedestrian of coring feature and stochastic subspace discrimination method again.
Background technology
Pedestrian recognizes again to be referred in the system of a multiple-camera composition, to the pedestrian image under different cameras visual angle
The problem of being matched.It provides key help for the analysis of the different aspects such as pedestrian's identity, behavior, and develops into
The key components of field of intelligent video surveillance.
Method main in identification field can be divided into two categories below to pedestrian again：1) pedestrian that feature based is represented recognizes again
Method；2) discrimination method again of the pedestrian based on metric learning.
In the pedestrian that feature based is represented again discrimination method, low-level visual feature is the most frequently used feature.Conventional is low
Layer visual signature has color histogram, texture etc., is described in detail below：Color histogram by distribution of color on statistical picture come
Whole image or the Color Distribution Features of one of zonule are described.It for visual angle change more robust, but easily by
The influence of the luminance transformations such as illumination, therefore its extraction generally on specific color space.Whole image of textural characteristics description
Or the structural information of one of zonule, it is a good supplement to the colouring information that color histogram feature is described.Greatly
Identification algorithm is all based on the realization of bottom visual signature to part pedestrian again, but works as mankind itself and recognize task again in progress pedestrian
When, be frequently not completely by bottom visual signature, it is but more by judging two images with semantic attributive character
Whether same a group traveling together is belonged to.Such as：Hair style, the type sympathized, the type of overcoat, the information such as shoes.Compared with bottom visual signature,
Method based on semantic attribute feature has natural advantage：Semantic attribute is for pedestrian's macroscopic features under different monitoring video
Difference more robust, same pedestrian is under different monitoring video, and the description of its semantic attribute is typically constant；Semantic attribute
Understanding with the mankind is more closely, therefore, the result that the characterization method based on semantic attribute is obtained more meets Man's Demands；It is based on
The method of semantic attribute is more convenient the interaction of people.
After character representation method, the distance for how measuring different pedestrian images is also the key that pedestrian recognizes field again
One of problem.The method of feature based is when calculating characteristic vector similitude, generally using Euclidean distance, COS distance and geodetic
The classical distance function such as distance.These classical distance functions do not consider that the characteristic therefore performance of sample are often bad.In recent years
Come, lot of documents use distance measure method, by mark sample training, obtain one more meet sample properties away from
From function, so as to improve performance.These methods are realized by learning the distance function of a geneva form.Wherein based on simple
And discrimination method is in prostatitis to the pedestrian of direct strategy (KISS) in performance again.However, this method is built upon sample point
On the theory hypothesis basis of cloth Gaussian distributed, but sample in reality not only will not ideally Gaussian distributed, very
To being possible to seriously deviate, so as to cause hydraulic performance decline.In addition, in a practical situation, sample size is often much smaller than feature
Dimension, so that causing the calculating of mahalanobis distance in metric learning becomes difficult even intangibility.
The content of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide one kind is based on coring feature
The integrated pedestrian discrimination method again with stochastic subspace, enabling the feature of extraction, more approx Gaussian distributed reconciles sample
The contradiction of scale and intrinsic dimensionality, it is to avoid SSS (small-scale sample) problem, so that improving performance.
The purpose of the present invention can be achieved through the following technical solutions：
A kind of pedestrian integrated based on coring feature and stochastic subspace discrimination method again, comprises the following steps：
S1, obtains the training sample set and test sample collection of pedestrian image, determines the coring function between two samples, coring
The output valve of function is one-dimensional real number, and in each sample set, same pedestrian has multiple images；Concentrated in training sample,
The corresponding relation of each pedestrian and its multiple image is known, is concentrated in test sample, is unknown；
S2, is converted into coring feature, the dimension of coring feature is training sample by the primitive character of two sample sets respectively
The number of samples of this concentration；
S3, in the coring feature space of training sample set, randomly selects multiple different subspaces, empty in each son respectively
Between under, calculate the covariance matrix and its inverse matrix of the coring feature difference of different pedestrian images pair, calculate identical pedestrian image
To coring feature difference covariance matrix and its inverse matrix, obtain the distribution function of the coring feature difference of image pair；Make
It is to learn a suitable distribution function by training sample with the purpose of training sample；
S4, respectively under each subspace (subspace is the subspace chosen in step S3 herein), calculates test sample collection
The difference of the coring feature of middle sample pair, according to difference covariance matrix and its inverse matrix and distribution function, calculate sample to for
The probability of identical pedestrian and the probability for different pedestrians, and it regard the ratio of two probability as the distance between two samples；
S5, carries out integrated to the distance that is calculated in variant subspace, obtains test sample and concentrates between this pair of various kinds
Final distance, for pedestrian's identification, finally apart from smaller, sample is higher to the possibility for identical pedestrian.It can so pick out
Test sample concentrates the image for belonging to same pedestrian.
In described step S1, coring function is gaussian kernel function, and the distribution function that step S3 is obtained is Gaussian Profile letter
Number.
In described step S1, coring function is k (xi,xj),Wherein, σ=1, xi、
xjI-th, j training sample is represented respectively.
In described step S2, the detailed process that primitive character is converted into coring feature includes：
Training set X is converted into coring feature
Wherein, m is training set number of samples, X ∈ Rd×m, d is sample characteristics dimension,
Test set Z is converted into coring feature
Wherein, n is test set number of samples,
In described step S3, the covariance Σ of the coring feature difference of different pedestrian images pair0Calculating formula is：
Wherein, yij=0 represents all samples pair for being not belonging to same a group traveling together, N0To meet the total sample number of the condition；
The covariance matrix Σ of the coring feature difference of identical pedestrian image pair1Calculating formula is：
Wherein, yij=1 represents all samples pair for being not belonging to same a group traveling together, N1To meet the total sample number of the condition.
In described step S4, between two samples is apart from calculating formula：
Compared with prior art, the present invention has advantages below：
(1) by primitive character be converted into difference distribution between coring feature, feature pair will closer to like Gaussian distributed,
This is the basic theory hypothesis of the learning distance metric of direct strategy based on simple；
(2) in non-linear space, coring feature often has stronger identification capability；
(3) complicated eigenvector projection is calculated into distance respectively into the less subspace of multiple dimensions, with random choosing
The mode of the subspace projection taken replaces traditional apart from calculation, and algorithm performance can be made to have obvious lifting, optimized
The process that sample distance is calculated, saves the expense of matrix operation；
(4) some subspace number of dimensions far smaller than sample size randomly selected, has reconciled sample in practical application
Scale is far smaller than the contradiction of the feature dimensions number of degrees, and then causes being accurately calculated for distance, efficiently.
Brief description of the drawings
Fig. 1 is the flow chart of the inventive method；
Fig. 2 (a) -2 (d) is the probability distribution graph of the feature difference between the sample pair before and after using the present embodiment method, its
In：2 (a) is the difference distribution of primitive character similarly hereinafter a group traveling together's sample pair；2 (b) is different pedestrian samples pair under primitive character
Difference distribution；2 (c) is the difference distribution of coring feature similarly hereinafter a group traveling together's sample pair；2 (d) is different pedestrian's samples under coring feature
This to difference distribution.
Fig. 3 (a), 3 (b) be the present embodiment method in the case of using different parameters and different characteristic in VIPeR (P=
316) pedestrian recognizes the CMC curves on public data collection again, wherein：3 (a) is bent using the CMC of different parameters under LOMO features
Line；3 (b) is the CMC curves under kCCA features using different parameters.
Embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.The present embodiment is with technical solution of the present invention
Premised on implemented, give detailed embodiment and specific operating process, but protection scope of the present invention is not limited to
Following embodiments.
Embodiment
A kind of pedestrian integrated based on coring feature and stochastic subspace discrimination method again, comprises the following steps：
Step one：Primitive character is converted into coring character representation, is described in detail below：
Obtain the training set X ∈ R of pedestrian imaged×mWith test set Z ∈ Rd×n, wherein sample characteristics dimension is d, training set
Number of samples is m, and test set number of samples is n, and uses xiI-th of training sample is represented, z is usediRepresent i-th of test sample.Make
Use coring functionTraining set X is converted into coring featureTest set Z is converted into coring
FeatureWherein σ=1.Specific transfer process is represented by
Step 2：In coring feature space, L different subspaces are randomly selected.It is described in detail below：Complete step
After one, the dimension of coring feature space is identical with the sample number of training set, i.e. m dimensions.L is randomly selected in this m dimensions individual different
Subspace Dk(k=1,2 ..., L), the dimension of subspace is
Step 3：Respectively under different subspaces, the covariance square of the feature difference between different pedestrian images pair is calculated
Battle array Σ0, and obtain inverse matrixCalculate the covariance matrix Σ of the feature difference between identical pedestrian image pair1, and obtain inverse square
Battle arrayIt is described in detail below：Respectively under different subspaces, the covariance of the feature difference between different pedestrian images pair is calculated
Matrix Σ0, specific calculation is
Wherein yij=0 represents all samples pair for being not belonging to same a group traveling together, N0To meet the total sample number of the condition.Meter
Calculate the covariance matrix covariance Σ of the feature difference between identical pedestrian image pair1, Σ1Specific calculation be
Wherein yij=1 represents all samples pair for belonging to identical pedestrian, N1To meet the total sample number of the condition.
Step 4：Respectively under different subspaces, based on gauss of distribution function and two inverse matrixsMeter
The difference of two features of calculation belongs to the probability of identical pedestrian and the difference of two features belongs to the probability of different pedestrians, and two
The ratio of probability regards the distance between sample as, is described in detail below：WithRepresent test sample under subspaceWithDifference,
Use H0Represent that i-th of test sample and j-th of test sample belong to different pedestrians' it is assumed that using H1Represent i-th test sample and
J-th of test sample belongs to the hypothesis of identical pedestrian.When the difference Gaussian distributed between sample pair, two can be respectively obtained
Difference is under individual assumed conditionProbability
WithRepresentThe logarithm of the ratio of the probability of two hypothesis is obeyed, can be obtained
According to Bayesian formula, formula (5) is convertible into
I.e.
Constant is removed, the distance under k-th of subspace between i-th of test sample and j-th of test sample can be obtained, represented
For
Step 5：The distance that is calculated in L different subspace is carried out integrated, obtain final distance, specifically describe
It is as follows：For the L distance obtained in step 4, being adjusted the distance using weighted mean method, progress is integrated, and final range formula can
To be expressed as：
As shown in figure 1, being the flow chart of the present embodiment, embodiment is as follows：
1) kernel function is determined；
2) primitive character of training sample is converted into coring feature；
3) primitive character of test sample is converted into coring feature；
4) in order to judge ziAnd zjWhether belong to same a group traveling together, use H0Represent that they are dissimilar, that is, be not belonging to same a group traveling together,
Use H1Represent that they are similar, that is, belong to same a group traveling together；
5) L sub-spaces D is randomly selected in coring feature spacek(k=1,2 ..., L)；
6) respectively in different subspaces, the covariance matrix Σ of the feature difference between different pedestrian images pair is calculated0,
And obtain inverse matrix
7) respectively in different subspaces, the covariance matrix Σ of the feature difference between identical pedestrian image pair is calculated1,
And obtain inverse matrix
8) respectively in different subspaces, feature difference is calculatedThe probability-distribution function of two kinds of hypothesis is obeyed, and will
The logarithmic function of probability ratioIt is used as the distance between sample；
9) respectively in different subspaces, probability ratio distance is converted into mahalanobis distance dk(zi,zj)；
10) distance that the middle calculating of L sub-spaces is obtained is integrated in, final sample distance is obtained.
Fig. 2 (a) -2 (d) is the probability distribution graph of the feature difference between the sample pair before and after using the present embodiment method, post
Shape figure is actual probability distribution, and linear graph is the Gaussian distribution curve drawn according to data variance, the four original spies compared
Respectively LOMO, kCCA, SCNCD, ELF18 are levied, they are pedestrian's widely used features in discrimination method again, its
In：2 (a) is the difference distribution of primitive character similarly hereinafter a group traveling together's sample pair；2 (b) is different pedestrian samples pair under primitive character
Difference distribution；2 (c) is the difference distribution of coring feature similarly hereinafter a group traveling together's sample pair；2 (d) is different pedestrian's samples under coring feature
This to difference distribution.
Fig. 3 (a), 3 (b) be the present embodiment method in the case of using different parameters and different characteristic in VIPeR (P=
316) pedestrian recognizes the rank-1 matching rates on public data collection again, and is compared with traditional regularization method.Wherein：3
(a) it is the rank-1 matching rates under LOMO features using different parameters；3 (b) is the rank- under kCCA features using different parameters
1 matching rate.
Table 1
Table 1 is that the present embodiment method is recognized openly again with other based on metric learning algorithm in VIPeR (P=316) pedestrian
Performance comparision on data set.
Table 2
Table 2 is that the present embodiment method is recognized again with other based on metric learning algorithm in PRID 450S (P=225) pedestrian
Performance comparision on public data collection.
Table 3
| KRKISS | NFST | MLAPG | XQDA | MFA | kLFDA | KISS | LFDA | |
| Time | 5.04 | 2.48 | 40.9 | 3.86 | 2.58 | 2.74 | 7.41 | 229.3 |
Table 3 is that the present embodiment method compares with other training time expenses based on metric learning algorithm.
Claims (6)
1. a kind of pedestrian integrated based on coring feature and stochastic subspace discrimination method again, it is characterised in that including following step
Suddenly：
S1, obtains the training sample set and test sample collection of pedestrian image, the coring function between two samples is determined, in each sample
This concentration, same pedestrian has multiple images；
S2, is converted into coring feature, the dimension of coring feature is training sample set by the primitive character of two sample sets respectively
In number of samples；
S3, in the coring feature space of training sample set, randomly selects multiple different subspaces, respectively in each subspace
Under, the covariance matrix and its inverse matrix of the coring feature difference of different pedestrian images pair are calculated, identical pedestrian image pair is calculated
Coring feature difference covariance matrix and its inverse matrix, obtain the distribution function of the coring feature difference of image pair；
S4, respectively under each subspace, calculates the difference that test sample concentrates the coring feature of sample pair, according to difference covariance
Matrix and its inverse matrix and distribution function, calculate sample to the probability for identical pedestrian and be different pedestrians probability, and by two
The ratio of individual probability is used as the distance between two samples；
S5, carries out integrated to the distance that is calculated in variant subspace, obtains test sample and concentrates final between this pair of various kinds
Distance, for pedestrian's identification, finally apart from smaller, sample is higher to the possibility for identical pedestrian.
2. a kind of pedestrian integrated based on coring feature and stochastic subspace according to claim 1 discrimination method again, its
It is characterised by, in described step S1, coring function is gaussian kernel function, and the distribution function that step S3 is obtained is Gaussian Profile letter
Number.
3. a kind of pedestrian integrated based on coring feature and stochastic subspace according to claim 1 or 2 discrimination method again,
Characterized in that, in described step S1, coring function is k (xi,xj),Wherein, σ=
1, xi、xjI-th, j training sample is represented respectively.
4. a kind of pedestrian integrated based on coring feature and stochastic subspace according to claim 3 discrimination method again, its
It is characterised by, in described step S2, the detailed process that primitive character is converted into coring feature includes：
Training set X is converted into coring feature
<mrow>
<mover>
<mi>X</mi>
<mo>~</mo>
</mover>
<mo>=</mo>
<mo>&lsqb;</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mn>1</mn>
</msub>
<mo>,</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mn>2</mn>
</msub>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mi>m</mi>
</msub>
<mo>&rsqb;</mo>
<mo>=</mo>
<mo>&lsqb;</mo>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mn>1</mn>
</msub>
<mo>,</mo>
<mi>X</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mn>2</mn>
</msub>
<mo>,</mo>
<mi>X</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>m</mi>
</msub>
<mo>,</mo>
<mi>X</mi>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
<mo>&Element;</mo>
<msup>
<mi>R</mi>
<mrow>
<mi>m</mi>
<mo>&times;</mo>
<mi>m</mi>
</mrow>
</msup>
<mo>,</mo>
</mrow>
Wherein, m is training set number of samples, X ∈ Rd×m, d is sample characteristics dimension,
<mrow>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mi>i</mi>
</msub>
<mo>=</mo>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<mi>X</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mrow>
<mo>&lsqb;</mo>
<mi>k</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mn>1</mn>
</msub>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mi>k</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mn>2</mn>
</msub>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<mi>k</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>m</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
</mrow>
<mi>T</mi>
</msup>
<mo>&Element;</mo>
<msup>
<mi>R</mi>
<mrow>
<mi>m</mi>
<mo>&times;</mo>
<mn>1</mn>
</mrow>
</msup>
<mo>,</mo>
</mrow>
Test set Z is converted into coring feature
<mrow>
<mover>
<mi>Z</mi>
<mo>~</mo>
</mover>
<mo>=</mo>
<mo>&lsqb;</mo>
<msub>
<mover>
<mi>z</mi>
<mo>~</mo>
</mover>
<mn>1</mn>
</msub>
<mo>,</mo>
<msub>
<mover>
<mi>z</mi>
<mo>~</mo>
</mover>
<mn>2</mn>
</msub>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<msub>
<mover>
<mi>z</mi>
<mo>~</mo>
</mover>
<mi>n</mi>
</msub>
<mo>&rsqb;</mo>
<mo>=</mo>
<mo>&lsqb;</mo>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>z</mi>
<mn>1</mn>
</msub>
<mo>,</mo>
<mi>X</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>z</mi>
<mn>2</mn>
</msub>
<mo>,</mo>
<mi>X</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>z</mi>
<mi>n</mi>
</msub>
<mo>,</mo>
<mi>X</mi>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
<mo>&Element;</mo>
<msup>
<mi>R</mi>
<mrow>
<mi>m</mi>
<mo>&times;</mo>
<mi>n</mi>
</mrow>
</msup>
<mo>,</mo>
</mrow>
Wherein, n is test set number of samples,
5. a kind of pedestrian integrated based on coring feature and stochastic subspace according to claim 4 discrimination method again, its
It is characterised by, in described step S3, the covariance matrix Σ of the coring feature difference of different pedestrian images pair0Calculating formula is：
<mrow>
<msub>
<mi>&Sigma;</mi>
<mn>0</mn>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msub>
<mi>N</mi>
<mn>0</mn>
</msub>
</mfrac>
<munder>
<mo>&Sigma;</mo>
<mrow>
<msub>
<mi>y</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<mo>=</mo>
<mn>0</mn>
</mrow>
</munder>
<mrow>
<mo>(</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mi>i</mi>
</msub>
<mo>-</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mi>i</mi>
</msub>
<mo>-</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mi>T</mi>
</msup>
</mrow>
Wherein, yij=0 represents all samples pair for being not belonging to same a group traveling together, N0To meet the total sample number of the condition；
The covariance matrix Σ of the coring feature difference of identical pedestrian image pair1Calculating formula is：
<mrow>
<msub>
<mi>&Sigma;</mi>
<mn>1</mn>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msub>
<mi>N</mi>
<mn>1</mn>
</msub>
</mfrac>
<munder>
<mo>&Sigma;</mo>
<mrow>
<msub>
<mi>y</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<mo>=</mo>
<mn>1</mn>
</mrow>
</munder>
<mrow>
<mo>(</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mi>i</mi>
</msub>
<mo>-</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mi>i</mi>
</msub>
<mo>-</mo>
<msub>
<mover>
<mi>x</mi>
<mo>~</mo>
</mover>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mi>T</mi>
</msup>
</mrow>
Wherein, yij=1 represents all samples pair for being not belonging to same a group traveling together, N1To meet the total sample number of the condition.
6. a kind of pedestrian integrated based on coring feature and stochastic subspace according to claim 5 discrimination method again, its
It is characterised by, in described step S4, between two samples is apart from calculating formula：
<mrow>
<mi>d</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>z</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>z</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mover>
<mi>z</mi>
<mo>~</mo>
</mover>
<mi>i</mi>
</msub>
<mo>-</mo>
<msub>
<mover>
<mi>z</mi>
<mo>~</mo>
</mover>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mi>T</mi>
</msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>&Sigma;</mi>
<mn>1</mn>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>&Sigma;</mi>
<mn>0</mn>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mo>)</mo>
</mrow>
<mrow>
<mo>(</mo>
<msub>
<mover>
<mi>z</mi>
<mo>~</mo>
</mover>
<mi>i</mi>
</msub>
<mo>-</mo>
<msub>
<mover>
<mi>z</mi>
<mo>~</mo>
</mover>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>.</mo>
</mrow>
2
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710212251.XA CN107122795B (en) | 2017-04-01 | 2017-04-01 | Pedestrian re-identification method based on coring characteristics and random subspace integration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710212251.XA CN107122795B (en) | 2017-04-01 | 2017-04-01 | Pedestrian re-identification method based on coring characteristics and random subspace integration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107122795A true CN107122795A (en) | 2017-09-01 |
| CN107122795B CN107122795B (en) | 2020-06-02 |
Family
ID=59724629
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710212251.XA Expired - Fee Related CN107122795B (en) | 2017-04-01 | 2017-04-01 | Pedestrian re-identification method based on coring characteristics and random subspace integration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107122795B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107909049A (en) * | 2017-11-29 | 2018-04-13 | 广州大学 | Pedestrian's recognition methods again based on least square discriminant analysis distance study |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103295242A (en) * | 2013-06-18 | 2013-09-11 | 南京信息工程大学 | Multi-feature united sparse represented target tracking method |
| CN103885049A (en) * | 2014-03-06 | 2014-06-25 | 西安电子科技大学 | Meter-wave radar low elevation estimating method based on minimum redundancy linear sparse submatrix |
| CN104008380A (en) * | 2014-06-16 | 2014-08-27 | 武汉大学 | Pedestrian detection method and system based on salient regions |
| CN104408705A (en) * | 2014-09-23 | 2015-03-11 | 西安电子科技大学 | Anomaly detection method of hyperspectral image |
| CN104616319A (en) * | 2015-01-28 | 2015-05-13 | 南京信息工程大学 | Multi-feature selection target tracking method based on support vector machine |
| WO2016026370A1 (en) * | 2014-08-22 | 2016-02-25 | Zhejiang Shenghui Lighting Co., Ltd. | High-speed automatic multi-object tracking method and system with kernelized correlation filters |
-
2017
- 2017-04-01 CN CN201710212251.XA patent/CN107122795B/en not_active Expired - Fee Related
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103295242A (en) * | 2013-06-18 | 2013-09-11 | 南京信息工程大学 | Multi-feature united sparse represented target tracking method |
| CN103885049A (en) * | 2014-03-06 | 2014-06-25 | 西安电子科技大学 | Meter-wave radar low elevation estimating method based on minimum redundancy linear sparse submatrix |
| CN104008380A (en) * | 2014-06-16 | 2014-08-27 | 武汉大学 | Pedestrian detection method and system based on salient regions |
| WO2016026370A1 (en) * | 2014-08-22 | 2016-02-25 | Zhejiang Shenghui Lighting Co., Ltd. | High-speed automatic multi-object tracking method and system with kernelized correlation filters |
| CN104408705A (en) * | 2014-09-23 | 2015-03-11 | 西安电子科技大学 | Anomaly detection method of hyperspectral image |
| CN104616319A (en) * | 2015-01-28 | 2015-05-13 | 南京信息工程大学 | Multi-feature selection target tracking method based on support vector machine |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107909049A (en) * | 2017-11-29 | 2018-04-13 | 广州大学 | Pedestrian's recognition methods again based on least square discriminant analysis distance study |
| CN107909049B (en) * | 2017-11-29 | 2020-07-31 | 广州大学 | Pedestrian re-identification method based on least square discriminant analysis distance learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107122795B (en) | 2020-06-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110443143B (en) | Multi-branch convolutional neural network fused remote sensing image scene classification method | |
| Oh et al. | Approaching the computational color constancy as a classification problem through deep learning | |
| CN102609681B (en) | Face recognition method based on dictionary learning models | |
| CN105512684B (en) | Logo automatic identifying method based on principal component analysis convolutional neural networks | |
| CN112800876B (en) | Super-spherical feature embedding method and system for re-identification | |
| CN104966085B (en) | A kind of remote sensing images region of interest area detecting method based on the fusion of more notable features | |
| CN109101938B (en) | Multi-label age estimation method based on convolutional neural network | |
| CN108876796A (en) | A kind of lane segmentation system and method based on full convolutional neural networks and condition random field | |
| CN108345850A (en) | The scene text detection method of the territorial classification of stroke feature transformation and deep learning based on super-pixel | |
| CN104268593A (en) | Multiple-sparse-representation face recognition method for solving small sample size problem | |
| CN103020265B (en) | The method and system of image retrieval | |
| Isobe et al. | Deep convolutional encoder-decoder network with model uncertainty for semantic segmentation | |
| CN106897669A (en) | A kind of pedestrian based on consistent iteration various visual angles transfer learning discrimination method again | |
| CN105740915A (en) | Cooperation segmentation method fusing perception information | |
| CN113723492A (en) | Hyperspectral image semi-supervised classification method and device for improving active deep learning | |
| CN104408731B (en) | Region graph and statistic similarity coding-based SAR (synthetic aperture radar) image segmentation method | |
| Yang et al. | Log-euclidean kernel-based joint sparse representation for hyperspectral image classification | |
| CN112329771B (en) | Deep learning-based building material sample identification method | |
| CN109614866A (en) | Method for detecting human face based on cascade deep convolutional neural networks | |
| CN110516533A (en) | A kind of pedestrian based on depth measure discrimination method again | |
| CN109635726A (en) | A kind of landslide identification method based on the symmetrical multiple dimensioned pond of depth network integration | |
| CN106874862A (en) | People counting method based on submodule technology and semi-supervised learning | |
| CN107657276B (en) | Weak supervision semantic segmentation method based on searching semantic class clusters | |
| CN114511012A (en) | SAR image and optical image matching method based on feature matching and position matching | |
| CN107392211A (en) | The well-marked target detection method of the sparse cognition of view-based access control model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200602 |