CN109684493A - A kind of image rearrangement sequence method based on k neighborhood distribution score - Google Patents
A kind of image rearrangement sequence method based on k neighborhood distribution score Download PDFInfo
- Publication number
- CN109684493A CN109684493A CN201910009038.8A CN201910009038A CN109684493A CN 109684493 A CN109684493 A CN 109684493A CN 201910009038 A CN201910009038 A CN 201910009038A CN 109684493 A CN109684493 A CN 109684493A
- Authority
- CN
- China
- Prior art keywords
- score
- image
- new
- initial
- reference picture
- 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
Abstract
The invention discloses a kind of image rearrangement sequence methods based on k neighborhood distribution score, on the basis of carrying out the initial sorted lists of drop sequence according to image similarity, resequence further according to the height of k neighborhood distribution score；Including obtaining initial sorted listsEstablish the expanding query collection Q (p, k) of test image p；By initial sorted lists Ω (p, G) as new test chart image set；P is added in initial reference image collection G, to obtain new reference picture collection Gnew={ p }+G；Corresponding temporary order list is calculated for each new test image；K sorting position Ls of the image in each temporary order list in search extension query set Q (p, k)；Calculate the following score of each reference picture in initial sorted lists Ω (p, G): position score Sd‑l, Overlap score Sd‑oAnd discrete score Sd‑d；Construct the field the k distribution score S of each reference picture in each initial sorted lists Ω (p, G)d；It is resequenced according to the field the k distribution score of each reference picture in initial sorted lists Ω (p, G).The present invention can reduce interference of the negative sample to sequence, improve the accuracy rate of image retrieval.
Description
Technical field
The present invention relates to field of image search, especially a kind of method for being ranked up to image.
Background technique
Image retrieval is based primarily upon image similarity judgement: concentrating each ginseng by calculating testing image and reference image data
The similarity of image is examined, is then ranked up according to the height of similarity, it usually will be with the highest ginseng of testing image similarity
Image is examined as top-1, the accuracy rate of top-1 plays the role of the accuracy rate of image searching result vital.But by
Not only there is positive sample in reference image data concentration, but also there are negative sample, negative sample, which calculates image similarity, has interference, by
In positive sample there are photo angle, block the problems such as, this negative sample and testing image for will cause certain angles or not blocking
Similarity be higher than positive sample and testing image similarity, then using in the prior art merely rely on similarity calculation obtain
Sequence inaccuracy, or even there is the case where negative sample is discharged to top-1.
Summary of the invention
In view of the above shortcomings of the prior art, the present invention provides a kind of image rearrangement sequence side based on k neighborhood distribution score
Method solves the simple image similarity that relies on and is ranked up the technical issues of being easy by negative sample interference, can reduce negative sample
Interference to sequence improves the accuracy rate of image retrieval.
In order to solve the above-mentioned technical problem, present invention employs the following technical solutions: one kind being distributed score based on k neighborhood
Image rearrangement sequence method, according to image similarity carry out drop sequence initial sorted lists on the basis of, further according to k neighborhood
The height of distribution score is resequenced；The following steps are included:
Step 1: obtaining the initial reference image collection G={ g comprising N reference picturesi| i=1,2,3 ..., N }, and count
The similarity for calculating individual test image p Yu each reference picture is ranked up from high to low according to image similarity, to obtain just
Beginning sorted listsWherein,It is the ginseng that i-th bit is come in initial sorted lists
Examine image；
Step 2: the expanding query collection Q (p, k) of test image p is established, as follows:
Q (p, k)={ p }+N (p, k-1)；
Wherein, p is test image, and preceding k-1 of the N (p, k-1) in initial sorted lists Ω (p, G) with reference to figure
Picture,
Step 3: by initial sorted listsAs new test chart image setAlso,
Step 4: test image p being added in initial reference image collection G, to obtain new reference picture collection Gnew=
{p}+G；
Step 5: traversing new test chart image set Pnew, calculate with PnewIn each image successively as new test image to new
Reference picture collection GnewIn each reference picture similarity, and according to similarity from high to low to new reference picture collection GnewIn
Each reference picture is ranked up, so that corresponding each new test image obtains temporary order list；
Step 6: according to temporary order list corresponding to each new test image, k in search extension query set Q (p, k)
Open sorting position L of the image in each temporary order list；
Step 7: calculating the following score of each reference picture in initial sorted lists Ω (p, G): position score Sd-l, overlapping
Score Sd-oAnd discrete score Sd-d；Wherein,Every score be calculated as follows respectively:
Wherein, in expanding query collection Q (p, k) test image p in new test imageCorresponding temporary order column
Sorting position in table is L (p)；Reference picture in expanding query collection Q (p, k)In new test imageCorresponding faces
When sorted lists in sorting position beThe weight coefficient of test image p isK-1 reference picturesWeight coefficient be
Wherein, N (p, k) indicates the nearest field k of test image p, i.e., preceding k ginsengs in initial sorted lists Ω (p, G)
Examine the image set of image composition；Indicate new test imageThe nearest field k, i.e., new test chart
PictureThe image set of preceding k reference picture compositions in corresponding temporary order list；N is sought in card [] expression
(p, k) withThe quantity of identical image；
Wherein, var [] indicates that variance calculates；
Step 8: the field k of each reference picture is distributed score S in each initial sorted lists Ω (p, G) of constructiond, by following
Any one mode:
Score: S is distributed using position score as the field kd=Sd-l；
Score: S is distributed using Overlap score as k neighborhoodd=Sd-o
Score: S is distributed using position score and the product of Overlap score as k neighborhoodd=Sd-l×Sd-o；
Score: S is distributed using discrete score and the product of Overlap score as k neighborhoodd=Sd-d×Sd-o；
Score: S is distributed using the product of position score, discrete score and Overlap score as k neighborhoodd=Sd-l×Sd-d×
Sd-o；
Step 9: score is distributed according to the field k of each reference picture in initial sorted lists Ω (p, G), it is from high to low, right
Each reference picture is resequenced in initial sorted lists Ω (p, G), to obtain being distributed reordering for score based on k neighborhood
List.
Preferably, the similarity judge index using characteristic distance as test image and reference picture.
Preferably, the average reference image quantity n that k value is possessed according to same target object in initial reference image collection G
It is determined, the value interval of k is [0.6n, 0.7n].
Compared with prior art, the invention has the following advantages:
1, the present invention is on the basis of carrying out the initial sorted lists of drop sequence with image similarity, then calculates each with reference to figure
Picture according to k neighborhood be distributed score, and according to k neighborhood be distributed score height resequence, thus overcome merely according to
Bad image similarity is ranked up, and not can avoid the defect of negative sample interference, be can reduce interference of the negative sample to sequence, is mentioned
The accuracy rate of hi-vision retrieval.
2,5 kinds of construction fields k provided by the invention are distributed score SdMode, can be on the basis of initial sorted lists
Improve accuracy rate.Wherein, especially work as Sd=Sd-l×Sd-d×Sd-oWhen, there is highest accuracy rate.
3, the present invention using characteristic distance be used as similarity judge index, as Euclidean distance in the prior art, cosine away from
From etc..
4, k value is determined according to the composition of initial reference image data set in the present invention, is counted according to many experiments, k
Value interval be [0.6n, 0.7n] when, highest accuracy rate can be obtained by reordering.
5, method for reordering of the invention calculates succinctly, and complexity is low, may be implemented efficiently to reorder.
Detailed description of the invention
Fig. 1 is the reference view of the corresponding temporary order list of each new test image in step 5；
Fig. 2 is the Computing Principle schematic diagram of Overlap score；
Fig. 3 is the effect diagram of the image rearrangement sequence method based on k neighborhood distribution score；
Fig. 4 is to change k value to the heavy recognition performance assessment figure of image data set CUHK03；
Fig. 5 is to change k value to the heavy recognition performance assessment figure of image data set Market1501.
Specific embodiment
The present invention is described in further detail with preferred embodiment with reference to the accompanying drawing.
A kind of image rearrangement sequence method based on k neighborhood distribution score is carrying out the initial of drop sequence with image similarity
On the basis of sorted lists, resequence further according to the height of k neighborhood distribution score；The following steps are included:
Step 1: obtaining the initial reference image collection G={ g comprising N reference picturesi| i=1,2,3 ..., N }, and count
The similarity for calculating individual test image p Yu each reference picture is ranked up from high to low according to image similarity, to obtain just
Beginning sorted listsWherein,It is the ginseng that i-th bit is come in initial sorted lists
Examine image.
Step 2: the expanding query collection Q (p, k) of test image p is established, as follows:
Q (p, k)={ p }+N (p, k-1)；
Wherein, p is test image, and preceding k-1 of the N (p, k-1) in initial sorted lists Ω (p, G) with reference to figure
Picture,
Step 3: by initial sorted listsAs new test chart image setAlso,
Step 4: test image p being added in initial reference image collection G, to obtain new reference picture collection Gnew=
{p}+G。
Step 5: traversing new test chart image set Pnew, calculate with PnewIn each image successively as new test image to new
Reference picture collection GnewIn each reference picture similarity, and according to similarity from high to low to new reference picture collection GnewIn
Each reference picture is ranked up, so that corresponding each new test image obtains temporary order list, refering to what is shown in Fig. 1, right
Answer each new test imageNew reference picture collection GnewIn each reference picture obtain new ranking, in Fig. 1 I indicate
Carry out new self-reference image set GnewBut it is not belonging to the image of the expanding query collection Q (p, k) of p.
Step 6: according to temporary order list corresponding to each new test image, k in search extension query set Q (p, k)
Sorting position L of the image in each temporary order list is opened, sorting position is the serial number in sorting, and the inverse of sorting position is
For sorting position score.
Step 7: calculating the following score of each reference picture in initial sorted lists Ω (p, G): position score Sd-l, overlapping
Score Sd-oAnd discrete score Sd-d；Wherein,Every score be calculated as follows respectively:
Wherein, in expanding query collection Q (p, k) test image p in new test imageCorresponding temporary order column
Sorting position in table is L (p)；Reference picture in expanding query collection Q (p, k)In new test imageCorresponding
Sorting position in temporary order list isThe weight coefficient of test image p isK-1 reference picturesWeight coefficient be
Wherein, the Computing Principle of Overlap score is with reference to shown in Fig. 2:
N (p, k) indicates the nearest field k of test image p, i.e., preceding k reference pictures in initial sorted lists Ω (p, G)
The image set of composition；Indicate new test imageThe nearest field k, i.e., new test imageThe image set of preceding k reference picture compositions in corresponding temporary order list；Card [] expression ask N (p,
K) withThe quantity of identical image；
Wherein, var [] indicates that variance calculates.
Step 8: the field k of each reference picture is distributed score S in each initial sorted lists Ω (p, G) of constructiond, by following
Any one mode:
Score: S is distributed using position score as the field kd=Sd-l；
Score: S is distributed using Overlap score as k neighborhoodd=Sd-o
Score: S is distributed using position score and the product of Overlap score as k neighborhoodd=Sd-l×Sd-o；
Score: S is distributed using discrete score and the product of Overlap score as k neighborhoodd=Sd-d×Sd-o；
Score: S is distributed using the product of position score, discrete score and Overlap score as k neighborhoodd=Sd-l×Sd-d×
Sd-o。
Step 9: score is distributed according to the field k of each reference picture in initial sorted lists Ω (p, G), it is from high to low, right
Each reference picture is resequenced in initial sorted lists Ω (p, G), to obtain being distributed reordering for score based on k neighborhood
List.
Its effect can refer to shown in Fig. 3: setting k=5；Top: test image p and preceding 9 samples in initial sequencing table,
Wherein N1-N2 is negative sample, and P1-P7 is the query expansion collection Q (p, 5) of positive sample test image p by p, N1, P1, P2 and N2 group
At it is new test image that N1, which is then arranged,.Intermediate: the nearest-neighbor N (N1,5) and Q (p, 5) of image N1 is in new list
In 5 location tags, be 39,1,504,437 and 47 respectively, obtain distribution and discrete score is respectively 0.141 and 4.11 × 10-
3.The number of Q (p, 5) and N (N1,5) identical image is 1, so overlapping is divided into 1, three kinds of scores are multiplied to obtain final score
0.58 × 10-3 seeks the score of remaining image using identical method.Bottom end: according to score height modification sequencing table, discovery
Positive sample P1, P2, P3, P4 and P5 are in the position of first five in final list.
It is big at two using characteristic distance as image similarity judge index further to verify effect of the invention
Itd is proposed method: CUHK03 and Market1501 is assessed on type data set.
CUHK03 is made of 13164 images, altogether 1467 pedestrians, is collected by two different cameras, including hand
The bounding box of dynamic mark and the bounding box detected by deformable part model (Deformable Part Model, DPM), at this
Single shot mode is used in text, data set can be divided into the test set of the training set comprising 1367 people and 100 people, image from second
Head selects image as test set, and each pedestrian randomly selects an image composition ginseng from the image at first camera visual angle
Examine image set.
Market1501 includes 32668 images of 1501 pedestrians from six video cameras, it is divided into two portions
Point: 19,732 images of 12,936 images from 751 pedestrians as training set and from 750 pedestrians are as test
Collection detects bounding box using DPM.Using the test protocol similar with CUHK03 data set.
(1) the image rearrangement sequence method that total evaluation is distributed based on the field k
In a specific embodiment, the image rearrangement sequence method based on the distribution of the field k of proposition is existing heavy with other
Sort method is compared, and will not use the recognition performance of any method for reordering as reference line, as shown in table 1:
Table 1
| Method for reordering | CUHK03 | Market1501 |
| Reference line | 91.2 | 82.9 |
| CDM | 91.5 | 83.3 |
| AQE | 91.3 | 83.1 |
| SCA | 92.0 | 83.5 |
| k-NN | 91.9 | 83.4 |
| k-reciprocal neighbors | 92.1 | 84.1 |
| K distribution is reordered | 93.5 | 85.7 |
By context dissimilarity measure (Contextual Dissimilarity Measure, CDM), average lookup extension
(Average Query Expansion, AQE), sparse context activation (Sparse Contextual Activation,
SCA), k- nearest-neighbor reorders (k-Nearest Neighbor re-ranking, k-NN) and k inverse encodes (k-
Reciprocal encoding) compared with method of the invention.Experimental result is as shown in table 1, the results showed that weight of the invention
Sort method may be implemented effective top-1 accuracy rate and improve, and the reference line of CUHK03 and Market1501 data set is respectively
91.2% and 82.9%, the value corresponding to two datasets k is 7 and 17, and 2.3% and 2.8% can be obtained by being reordered with k distribution
Promotion, it can be found that this strategy has been over other methods.
(2) the different configuration mode of the assessment field k distribution score
It is previously mentioned that the k distribution score that reorders can consist of three parts: position score, discrete score and being overlapped
Point, therefore these three independent scores and generated by its combination of two and obtain the potential score of the other three: " position score × discrete
Point ", " position score × Overlap score " and " discrete score × Overlap score " can form six kinds and obtain classifying type.With data set
For CUHK03, the experimental result of every kind of score is shown in Table 2.
Table 2
| Score type | Top-1 |
| Position score | 92.9 |
| Discrete score | 6.7 |
| Overlap score | 93.2 |
| Position score × discrete score | 84.9 |
| Position score × Overlap score | 92.6 |
| Discrete score × Overlap score | 92.9 |
| Position score × discrete score × Overlap score | 93.5 |
It can be found that the score comprising three parts reaches optimum efficiency, combined better than other scores, than the list of best performance
Seed type " Overlap score " and composite type " discrete score × Overlap score " difference high 0.3% and 0.6%.It is noticeable
It is that the effect based on " discrete score " is poor, can only achieve 6.7%, decline is more compared with reference line 91.2%, but gives up this
After item score, " position score × Overlap score " combination obtains 92.6% top-1 accuracy rate, than containing there are three types of the groups of score
Conjunction reduces 0.9%.Therefore, it is combined with " position score " and " Overlap score " to realize the complementary of different attribute information
And it finally obtains optimum performance and is promoted.
(3) influence of k value is assessed
Parameter influences: in the image rearrangement sequence method of the field k distribution score, k value be it is variable, what is presented before is all
In chart, corresponding to two datasets value is k=7 and k=17, it is contemplated that the composition of each data set is different, and assessment changes k
It is worth the influence of counterweight recognition performance.As shown in figs. 4 and 5, discovery is on both data sets when parameter k value is in section [6,14]
When within [9,23], performance is promoted to 0.8~2.3% for CUHK03 better than reference line, as k=7, reaches optimal
Top-1 accuracy rate is 93.5%, for Market1501, is promoted to 0.5~2.8%, can get 85.7% in k=17 most
In high precision.The reference picture concentration of two datasets averagely has 9.76 and 26.3 images with a group traveling together, it is seen that when the value of k is
At 0.6~0.7 times of the value, higher heavy recognition performance is may be implemented in the strategy that reorders of proposition.In addition, when k is more than certain
After threshold value, top-1 accuracy rate will be gradually decreased, and more negative samples will be introduced and increase error by being primarily due to biggish parameter k,
So that reducing performance and improving computation complexity.
In conclusion the image rearrangement sequence method of the invention based on k neighborhood distribution score, passes through construction k neighborhood distribution
Score selects to can be improved the score of positive sample suitably in k value, so that the interference of negative sample reorder is reduced, and it
Be it is unsupervised, can adapt to various image retrieval tasks.
Claims (3)
1. a kind of image rearrangement sequence method based on k neighborhood distribution score, it is characterised in that: carrying out drop row with image similarity
On the basis of the initial sorted lists of sequence, resequence further according to the height of k neighborhood distribution score；The following steps are included:
Step 1: obtaining the initial reference image collection G={ g comprising N reference picturesi| i=1,2,3 ..., N }, and calculate individual
The similarity of test image p and each reference picture are ranked up from high to low according to image similarity, to obtain initial sequence
ListWherein,Be come in initial sorted lists i-th bit with reference to figure
Picture；
Step 2: the expanding query collection Q (p, k) of test image p is established, as follows:
Q (p, k)={ p }+N (p, k-1)；
Wherein, p is test image, and preceding k-1 of the N (p, k-1) in initial sorted lists Ω (p, G) opens reference pictures,
Step 3: by initial sorted listsAs new test chart image setAlso,
Step 4: test image p being added in initial reference image collection G, to obtain new reference picture collection Gnew={ p }+
G；
Step 5: traversing new test chart image set Pnew, calculate with PnewIn each image successively as new test image to new ginseng
Examine image set GnewIn each reference picture similarity, and according to similarity from high to low to new reference picture collection GnewIn respectively join
It examines image to be ranked up, so that corresponding each new test image obtains temporary order list；
Step 6: according to temporary order list corresponding to each new test image, scheming for k in search extension query set Q (p, k)
As the sorting position L in each temporary order list；
Step 7: calculating the following score of each reference picture in initial sorted lists Ω (p, G): position score Sd-l, Overlap score
Sd-oAnd discrete score Sd-d；Wherein,Every score be calculated as follows respectively:
Wherein, in expanding query collection Q (p, k) test image p in new test imageIn corresponding temporary order list
Sorting position be L (p)；Reference picture in expanding query collection Q (p, k)In new test imageCorresponding is interim
Sorting position in sorted lists isThe weight coefficient of test image p isK-1 reference pictures
Weight coefficient be
Wherein, N (p, k) indicates the nearest field k of test image p, i.e., preceding k in initial sorted lists Ω (p, G) are with reference to figure
As the image set of composition；Indicate new test imageThe nearest field k, i.e., new test imageThe image set of preceding k reference picture compositions in corresponding temporary order list；Card [] expression ask N (p,
K) withThe quantity of identical image；
Wherein, var [] indicates that variance calculates；
Step 8: the field k of each reference picture is distributed score S in each initial sorted lists Ω (p, G) of constructiond, by following any
A kind of mode:
Score: S is distributed using position score as the field kd=Sd-l；
Score: S is distributed using Overlap score as k neighborhoodd=Sd-o
Score: S is distributed using position score and the product of Overlap score as k neighborhoodd=Sd-l×Sd-o；
Score: S is distributed using discrete score and the product of Overlap score as k neighborhoodd=Sd-d×Sd-o；
Score: S is distributed using the product of position score, discrete score and Overlap score as k neighborhoodd=Sd-l×Sd-d×Sd-o；
Step 9: score being distributed according to the field k of each reference picture in initial sorted lists Ω (p, G), from high to low, to initial
Each reference picture is resequenced in sorted lists Ω (p, G), to obtain the retracing sequence for being distributed score based on k neighborhood
Table.
2. the image rearrangement sequence method according to claim 1 based on k neighborhood distribution score, it is characterised in that: using special
Similarity judge index of the sign distance as test image and reference picture.
3. it is according to claim 1 based on k neighborhood distribution score image rearrangement sequence method, it is characterised in that: k value according to
The average reference image quantity n that same target object is possessed in initial reference image collection G is determined, and the value interval of k is
[0.6n,0.7n]。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910009038.8A CN109684493B (en) | 2019-01-04 | 2019-01-04 | Image reordering method based on k neighborhood distribution score |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910009038.8A CN109684493B (en) | 2019-01-04 | 2019-01-04 | Image reordering method based on k neighborhood distribution score |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109684493A true CN109684493A (en) | 2019-04-26 |
| CN109684493B CN109684493B (en) | 2020-06-23 |
Family
ID=66192524
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910009038.8A Active CN109684493B (en) | 2019-01-04 | 2019-01-04 | Image reordering method based on k neighborhood distribution score |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109684493B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110502964A (en) * | 2019-05-21 | 2019-11-26 | 杭州电子科技大学 | A kind of pedestrian's recognition methods again of unsupervised data-driven |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103325122A (en) * | 2013-07-03 | 2013-09-25 | 武汉大学 | Pedestrian retrieval method based on bidirectional sequencing |
| CN107506401A (en) * | 2017-08-02 | 2017-12-22 | 大连理工大学 | A kind of image retrieval rearrangement method based on drawing method |
| CN107506703A (en) * | 2017-08-09 | 2017-12-22 | 中国科学院大学 | A kind of pedestrian's recognition methods again for learning and reordering based on unsupervised Local Metric |
| US20180341810A1 (en) * | 2013-08-13 | 2018-11-29 | Logograb Limited | Recognition Process Of An Object In A Query Image |
-
2019
- 2019-01-04 CN CN201910009038.8A patent/CN109684493B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103325122A (en) * | 2013-07-03 | 2013-09-25 | 武汉大学 | Pedestrian retrieval method based on bidirectional sequencing |
| US20180341810A1 (en) * | 2013-08-13 | 2018-11-29 | Logograb Limited | Recognition Process Of An Object In A Query Image |
| CN107506401A (en) * | 2017-08-02 | 2017-12-22 | 大连理工大学 | A kind of image retrieval rearrangement method based on drawing method |
| CN107506703A (en) * | 2017-08-09 | 2017-12-22 | 中国科学院大学 | A kind of pedestrian's recognition methods again for learning and reordering based on unsupervised Local Metric |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110502964A (en) * | 2019-05-21 | 2019-11-26 | 杭州电子科技大学 | A kind of pedestrian's recognition methods again of unsupervised data-driven |
| CN110502964B (en) * | 2019-05-21 | 2021-09-28 | 杭州电子科技大学 | Unsupervised data-driven pedestrian re-identification method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109684493B (en) | 2020-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Fergus et al. | A visual category filter for google images | |
| US8892542B2 (en) | Contextual weighting and efficient re-ranking for vocabulary tree based image retrieval | |
| CN101187927B (en) | Criminal case joint investigation intelligent analysis method | |
| CN102388392B (en) | Pattern recognition device | |
| CN101770578B (en) | Image characteristic extraction method | |
| CN107103326A (en) | The collaboration conspicuousness detection method clustered based on super-pixel | |
| CN106469299A (en) | A kind of vehicle search method and device | |
| CN109063649B (en) | Pedestrian re-identification method based on twin pedestrian alignment residual error network | |
| CN102902826B (en) | A kind of image method for quickly retrieving based on reference picture index | |
| CN103810299A (en) | Image retrieval method on basis of multi-feature fusion | |
| CN103336795A (en) | Video indexing method based on multiple features | |
| CN107292259A (en) | The integrated approach of depth characteristic and traditional characteristic based on AdaRank | |
| CN107291936A (en) | The hypergraph hashing image retrieval of a kind of view-based access control model feature and sign label realizes that Lung neoplasm sign knows method for distinguishing | |
| CN113177612B (en) | Agricultural pest image identification method based on CNN few samples | |
| CN103810252A (en) | Image retrieval method based on group sparse feature selection | |
| CN104462550B (en) | Pedestrian's recognition methods again of sorting consistence is merged based on similitude and dissimilarity | |
| CN112307232A (en) | Intelligent classification storage processing method for big data content | |
| CN105678244B (en) | A kind of near video search method based on improved edit-distance | |
| CN103761503A (en) | Self-adaptive training sample selection method for relevance feedback image retrieval | |
| CN109934852B (en) | Video description method based on object attribute relation graph | |
| CN108121806A (en) | One kind is based on the matched image search method of local feature and system | |
| CN101515286A (en) | Image matching method based on image feature multi-level filtration | |
| CN109684493A (en) | A kind of image rearrangement sequence method based on k neighborhood distribution score | |
| CN110928986A (en) | Legal evidence sorting and recommending method, device, equipment and storage medium | |
| CN111125396A (en) | Image retrieval method of single-model multi-branch structure |
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 |