CN110287883A - A method of recognition of face is carried out based on nearest neighbor distance ratio method is improved - Google Patents
A method of recognition of face is carried out based on nearest neighbor distance ratio method is improved Download PDFInfo
- Publication number
- CN110287883A CN110287883A CN201910559456.4A CN201910559456A CN110287883A CN 110287883 A CN110287883 A CN 110287883A CN 201910559456 A CN201910559456 A CN 201910559456A CN 110287883 A CN110287883 A CN 110287883A
- Authority
- CN
- China
- Prior art keywords
- face picture
- feature descriptor
- face
- inquiry
- matched
- 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.)
- Pending
Links
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
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Human Computer Interaction (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Collating Specific Patterns (AREA)
Abstract
The present invention discloses a kind of method for carrying out recognition of face based on improvement nearest neighbor distance ratio method, is related to technical field of face recognition.This method extracts the key point of inquiry face picture and face picture to be matched using SIFT algorithm, feature descriptor corresponding with key point is generated simultaneously, arest neighbors is obtained by calculating the Euclidean distance in inquiry face picture between the feature descriptor and the feature descriptor of all key points in face picture to be matched of each key point to match pair, by searching for the to be matched face picture most with inquiry face picture match number, final recognition of face is completed.This method had not only remained the powerful judgement index of arest neighbors ratio method, but also can fully take into account time neighbour and the effect of other neighbours, and the registration rate of recognition of face can be improved.
Description
Technical field
The present invention relates to technical field of face recognition, specifically a kind of to be carried out based on improvement nearest neighbor distance ratio method
The method of recognition of face.
Background technique
Nowadays face recognition technology will be used wider and wider, and the main thinking of realization has two, first is that using figure
As registration technique, retrieve and the most picture of current image match point in the database;Second is that the method based on classification, uses volume
Product neural network directly classifies to picture.
SIFT feature is Scale invariant features transform, since it can keep not rotation, scaling, brightness change etc.
Deformation, and there is very strong real-time and increased enrollment, so being widely used in image registration field, especially face recognition technology
On.So we are also using SIFT feature descriptor as the character representation of image key points.
It obtains needing the descriptors match to picture key point after SIFT feature descriptor, matching strategy mainly has three
Kind, first is that the distance between threshold method, i.e. target critical point feature descriptor and inquiry key point feature descriptor are less than some
Threshold value is considered as successful match;Second is that nearest neighbor method, i.e., inquiry key point feature descriptor only matches and it is apart from nearest mesh
Mark key point is matched, only one best matching result of the method;Third is that arest neighbors ratio method, that is, pass through inquiry key point
It is true with the ratio between the distance between the distance between arest neighbors key point feature in Target Photo and target time neighbour's key point feature
Whether the arest neighbors key point that sets the goal can match with inquiry face picture key point, and the method, which is easier to find, more judgement index
Matching pair.
Now, recognition of face is an active research direction of machine learning and area of pattern recognition.And in industry
Boundary has had to be widely applied very much, video monitoring, file administration, in terms of effect it is significant.
But the matching strategy of the Face datection based on retrieval effect and pessimistic for various reasons, take current effect
For best arest neighbors ratio method, the ratio between arest neighbors and secondary neighbour only considered in matching, that is to say, that only close
It has infused the arest neighbors of the condition of satisfaction, and has tested and show in the case where being unsatisfactory for condition, correctly matched quantity is even by secondary neighbour
The quantity of third neighborhood matching is also very considerable, so traditional arest neighbors ratio method reduces correct matched feature points
Mesh causes the accuracy rate being finally registrated not high, and matching effect is also bad.
Summary of the invention
For the present invention in order to make up the deficiency of existing matching strategy, it is smaller to solve the matching range in recognition of face, ignores time
The effect of neighborhood matching point and other match points and the problem for causing matching effect undesirable provide a kind of based on improving arest neighbors
The method of distance ratio method progress recognition of face.
Of the invention is a kind of based on the method for improving the progress recognition of face of nearest neighbor distance ratio method, solves above-mentioned technology and asks
Topic the technical solution adopted is as follows:
A method of recognition of face being carried out based on nearest neighbor distance ratio method is improved, this method is extracted using SIFT algorithm
The key point of face picture and face picture to be matched is inquired, while generating feature descriptor corresponding with key point, is passed through
Calculate the feature descriptor and the feature of all key points in face picture to be matched of each key point in inquiry face picture
Euclidean distance between descriptor obtains arest neighbors matching pair, by searching for most to be matched of inquiry face picture match number
Face picture completes final recognition of face.
The specific implementation step of this method includes:
1) input inquiry face picture and face picture to be matched;
2) key point in inquiry face picture and face picture to be matched is extracted using SIFT algorithm, and to each key
Point generates the feature descriptor of one 128 dimension;
3) each feature descriptor and all feature descriptors in face picture to be matched in inquiry face picture are calculated
Between Euclidean distance, set Q is ranked up and is stored in the Euclidean distance being calculated according to ascending sequence, arrange
It is the sequence of feature descriptor in face picture to be matched by the sequence corresponding conversion of Euclidean distance in set Q after the completion of sequence;
4) setting pairing threshold value threshold, the feature descriptor for calculating inquiry face picture are adjacent with front and back in set Q
Euclidean distance between two feature descriptors, if meetingThen the feature for inquiring face picture is retouched
Symbol is stated with previous feature descriptor in set Q as neighborhood matching pair;
5) circulation executes step 4), obtains all neighborhood matchings pair of all feature descriptors in inquiry face picture, and
It is stored in set P;
6) uniqueness of the included neighborhood matching pair of set P is checked;
7) record queries face picture is with face picture to be matched with logarithm;
8) step 1)-step 7) is successively executed to inquiry face picture and all face pictures to be matched;
9) the to be matched face picture most with inquiry face picture match number is found, and determines the face picture to be matched
It is same face with inquiry face picture.
Optionally, involved face picture to be matched is transferred from database;
Before transferring face picture to be matched from database, first face pictures all in database tentatively can be sieved
Choosing:
Set minimum threshold;
The face picture to be matched for being greater than minimum threshold with the similarity of inquiry face picture is transferred in the database.
Specifically, when executing step 3),
Firstly, settingFor inquire face picture i-th of key point feature descriptor,For face picture to be matched
J-th of key point feature descriptor;
Then, each is calculatedWith it is allBetween Euclidean distance, according to ascending sequence to being calculated
Euclidean distance be ranked up and be stored in set Qi, after the completion of sequence, by the sequence corresponding conversion of Euclidean distance in set Qi
For the sequence of feature descriptor in face picture to be matched;
Finally,And withIn corresponding set Qi, adjustment set Qi puts in order, by set Qi andIt is denoted as apart from nearest feature descriptorSecond close feature descriptor is denoted asAnd so on be marked.
Optionally, the detailed process of step 4) is executed are as follows:
4-1) setting pairing threshold value threshold,
4-2) calculate the feature descriptor of inquiry face pictureWith adjacent two feature descriptor in front and back in set Qi Between Euclidean distance,
If meetingThe feature descriptor of face picture will then be inquiredBefore set Qi
One feature descriptorAs arest neighbors matching pair,
If being unsatisfactory forThen indicate the feature descriptor of inquiry face pictureWithout neighbour
Pairing;
4-3) calculate the feature descriptor of inquiry face pictureWith adjacent two feature descriptor in front and back in set Qi Between Euclidean distance,
If meetingThe feature descriptor of face picture will then be inquiredBefore set Qi
One feature descriptorAs secondary neighborhood matching pair;
If being unsatisfactory forThen indicate the feature descriptor of inquiry face pictureWithout neighbour
Pairing;
4-4) and so on, calculate the feature descriptor of inquiry face pictureIt is retouched with adjacent two feature in front and back in set Qi
The Euclidean distance between symbol is stated, the feature descriptor until completing inquiry face pictureIt is retouched with last adjacent two feature of set Qi
The Euclidean distance between symbol is stated, i.e., has found the feature descriptor with inquiry face picture in set QiIt is relevant all close
Neighbour's matching pair.
Specifically, after executing step 4),
Neighborhood matching relevant to same feature descriptor in inquiry face picture is to being stored in the same set;
From the same incoherent neighborhood matching of feature descriptor in inquiry face picture to being stored in different set;
All set comprising all feature descriptors in inquiry face picture are stored in set P.
In step 6), the uniqueness of the included neighborhood matching pair of set P, concrete operations are checked are as follows:
Check each key point of inquiry face picture corresponding thereto the key point in face picture to be matched whether
It matches, the matching pair of arest neighbors is retained in matched situation, complete the uniqueness inspection of set P.
After the uniqueness for checking the included neighborhood matching pair of set P, in inquiry face picture and face picture to be matched
All key point matchings are to using RANSAC algorithm to reject mispairing pair.
Of the invention is a kind of based on the method for improving the progress recognition of face of nearest neighbor distance ratio method, compared with prior art
It has the beneficial effect that
1) on the one hand, the present invention improves arest neighbors ratio method, carries out inquiry face picture and face figure to be matched
Matching between the key point of piece can significantly improve raising face picture a possibility that fully considering other neighborhood matchings
The recall rate of registration;
2) on the other hand, when carrying out the matching of key point, the powerful judgement index of arest neighbors ratio method had not only been remained, but also
Time neighbour and the effect of other neighbours can be fully taken into account, it is final to improve registration rate.
Detailed description of the invention
Attached drawing 1 is method schematic of the invention.
Specific embodiment
The technical issues of to make technical solution of the present invention, solving and technical effect are more clearly understood, below in conjunction with tool
Body embodiment carries out clear, complete description to technical solution of the present invention, it is clear that described embodiment is only this hair
Bright a part of the embodiment, instead of all the embodiments.Based on the embodiment of the present invention, those skilled in the art are not doing
All embodiments obtained under the premise of creative work out, all within protection scope of the present invention.
Embodiment one:
In conjunction with attached drawing 1, the present embodiment proposes a kind of method for carrying out recognition of face based on improvement nearest neighbor distance ratio method,
This method extracts the key point of inquiry face picture and face picture to be matched using SIFT algorithm, while generating and key point phase
Corresponding feature descriptor, by the feature descriptor and face figure to be matched that calculate each key point in inquiry face picture
Euclidean distance in piece between the feature descriptor of all key points obtain arest neighbors matching pair, by searching for inquiry face figure
The piece to be matched face picture most with logarithm, completes final recognition of face.
The specific implementation step of this method includes:
1) face picture to be matched, input inquiry face picture and the face picture to be matched transferred are transferred from database.
2) key point in inquiry face picture and face picture to be matched is extracted using SIFT algorithm, and to each key
Point generates the feature descriptor of one 128 dimension;IfFor inquire face picture i-th of key point feature descriptor,For
The feature descriptor of j-th of key point of face picture to be matched.
3) each is calculatedWith it is allBetween Euclidean distance, according to ascending sequence to being calculated
Euclidean distance is ranked up and is stored in set Qi, and after the completion of sequence, the sequence corresponding conversion by Euclidean distance in set Qi is
The sequence of feature descriptor in face picture to be matched;
Then, existAnd withIn corresponding set Qi, adjustment set Qi puts in order, by set Qi andIt is denoted as apart from nearest feature descriptorSecond close feature descriptor is denoted asAnd so on be marked.
4) setting pairing threshold value threshold,
A1 the feature descriptor of inquiry face picture) is calculatedWith adjacent two feature descriptor in front and back in set QiBetween Euclidean distance,
If meetingThe feature descriptor of face picture will then be inquiredBefore set Qi
One feature descriptorAs arest neighbors matching pair,
If being unsatisfactory forThen indicate the feature descriptor of inquiry face pictureWithout neighbour
Pairing;
A2 the feature descriptor of inquiry face picture) is calculatedWith adjacent two feature descriptor in front and back in set QiBetween Euclidean distance,
If meetingThe feature descriptor of face picture will then be inquiredBefore set Qi
One feature descriptorAs secondary neighborhood matching pair;
If being unsatisfactory forThen indicate the feature descriptor of inquiry face pictureWithout neighbour
Pairing;
A3) and so on, calculate the feature descriptor of inquiry face pictureIt is retouched with adjacent two feature in front and back in set Qi
The Euclidean distance between symbol is stated, the feature descriptor until completing inquiry face pictureIt is retouched with last adjacent two feature of set Qi
The Euclidean distance between symbol is stated, i.e., has found the feature descriptor with inquiry face picture in set QiIt is relevant all close
Neighbour's matching pair.
5) circulation executes step 4), obtains all neighborhood matchings pair of all feature descriptors in inquiry face picture, and
It is stored in set P.
6) uniqueness of the included neighborhood matching pair of set P, concrete operations are checked are as follows:
Check each key point of inquiry face picture corresponding thereto the key point in face picture to be matched whether
It matches, the matching pair of arest neighbors is retained in matched situation, complete the uniqueness inspection of set P.
7) record queries face picture is with face picture to be matched with logarithm.
8) step 1)-step 7) is successively executed to inquiry face picture and all face pictures to be matched.
9) the to be matched face picture most with inquiry face picture match number is found, and determines the face picture to be matched
It is same face with inquiry face picture.
It, can be first in database before transferring face picture to be matched from database in the step 1) of the present embodiment
All face pictures carry out preliminary screening:
Set minimum threshold;
The face picture to be matched for being greater than minimum threshold with the similarity of inquiry face picture is transferred in the database.
In the present embodiment, after executing step 4),
Neighborhood matching relevant to same feature descriptor in inquiry face picture is to being stored in the same set;
From the same incoherent neighborhood matching of feature descriptor in inquiry face picture to being stored in different set;
All set comprising all feature descriptors in inquiry face picture are stored in set P.
It in the present embodiment, i.e., can be with after the uniqueness of inspection the included neighborhood matching pair of set P after executing step 6)
Inquiry face picture is matched with key point all in face picture to be matched to using the rejecting mispairing pair of RANSAC algorithm.
In summary, using of the invention a kind of based on the method improved nearest neighbor distance ratio method and carry out recognition of face,
By SIFT algorithm extract inquiry face picture and face picture to be matched key point, and carry out inquirer's face picture and to
Key point matching with face picture, improves pairing accuracy rate.
Use above specific case elaborates the principle of the present invention and embodiment, these embodiments are
It is used to help understand core of the invention technology contents, the protection scope being not intended to restrict the invention, technical side of the invention
Case is not limited in above-mentioned specific embodiment.Based on above-mentioned specific embodiment of the invention, those skilled in the art
Without departing from the principle of the present invention, any improvement and modification to made by the present invention should all be fallen into of the invention special
Sharp protection scope.
Claims (8)
1. a kind of based on the method for improving the progress recognition of face of nearest neighbor distance ratio method, which is characterized in that this method uses
SIFT algorithm extracts the key point of inquiry face picture and face picture to be matched, while generating feature corresponding with key point
Descriptor, it is related with institute in face picture to be matched by the feature descriptor for calculating each key point in inquiry face picture
Euclidean distance between the feature descriptor of key point obtain arest neighbors matching pair, by searching for inquiry face picture match number most
More face pictures to be matched, completes final recognition of face.
2. it is according to claim 1 a kind of based on the method for improving the progress recognition of face of nearest neighbor distance ratio method, it is special
Sign is that the specific implementation step of this method includes:
1) input inquiry face picture and face picture to be matched;
2) key point in inquiry face picture and face picture to be matched is extracted using SIFT algorithm, and raw to each key point
The feature descriptor tieed up at one 128;
3) it calculates in inquiry face picture in each feature descriptor and face picture to be matched between all feature descriptors
Euclidean distance, set Q is ranked up and is stored in the Euclidean distance being calculated according to ascending sequence, has been sorted
The sequence corresponding conversion of Euclidean distance in set Q is the sequence of feature descriptor in face picture to be matched by Cheng Hou;
4) threshold value threshold is matched in setting, calculates adjacent two spy in front and back in the feature descriptor and set Q of inquiry face picture
The Euclidean distance between descriptor is levied, if meetingThe feature descriptor of face picture will then be inquired
With feature descriptor previous in set Q as neighborhood matching pair;
5) circulation executes step 4), obtains all neighborhood matchings pair of all feature descriptors in inquiry face picture, and store
In set P;
6) uniqueness of the included neighborhood matching pair of set P is checked;
7) record queries face picture is with face picture to be matched with logarithm;
8) step 1)-step 7) is successively executed to inquiry face picture and all face pictures to be matched;
9) the to be matched face picture most with inquiry face picture match number is found, and determines the face picture to be matched and looks into
Inquiry face picture is same face.
3. it is according to claim 1 a kind of based on the method for improving the progress recognition of face of nearest neighbor distance ratio method, it is special
Sign is that face picture to be matched is transferred from database;
Before transferring face picture to be matched from database, preliminary screening first can be carried out to face pictures all in database:
Set minimum threshold;
The face picture to be matched for being greater than minimum threshold with the similarity of inquiry face picture is transferred in the database.
4. it is according to claim 1 a kind of based on the method for improving the progress recognition of face of nearest neighbor distance ratio method, it is special
Sign is, when executing step 3),
Firstly, settingFor inquire face picture i-th of key point feature descriptor,For the jth of face picture to be matched
The feature descriptor of a key point;
Then, each is calculatedWith it is allBetween Euclidean distance, according to ascending sequence to the Europe being calculated
Family name's distance is ranked up and is stored in set Qi, after the completion of sequence, by the sequence corresponding conversion of Euclidean distance in set Qi be to
Match the sequence of feature descriptor in face picture;
Finally,And withIn corresponding set Qi, adjustment set Qi puts in order, by set Qi andAway from
It is denoted as from nearest feature descriptorSecond close feature descriptor is denoted asAnd so on be marked.
5. it is according to claim 4 a kind of based on the method for improving the progress recognition of face of nearest neighbor distance ratio method, it is special
Sign is, executes the detailed process of step 4) are as follows:
4-1) setting pairing threshold value threshold,
4-2) calculate the feature descriptor of inquiry face pictureWith adjacent two feature descriptor in front and back in set Qi
Between Euclidean distance,
If meetingThe feature descriptor of face picture will then be inquiredWith the previous spy of set Qi
Levy descriptorAs arest neighbors matching pair,
If being unsatisfactory forThen indicate the feature descriptor of inquiry face pictureWithout neighborhood matching pair;
4-3) calculate the feature descriptor of inquiry face pictureWith adjacent two feature descriptor in front and back in set Qi
Between Euclidean distance,
If meetingThe feature descriptor of face picture will then be inquiredWith the previous spy of set Qi
Levy descriptorAs secondary neighborhood matching pair;
If being unsatisfactory forThen indicate the feature descriptor of inquiry face pictureWithout neighborhood matching pair;
4-4) and so on, calculate the feature descriptor of inquiry face pictureWith adjacent two feature descriptor in front and back in set Qi
Between Euclidean distance, until complete inquiry face picture feature descriptorWith last adjacent two feature descriptor of set Qi
Between Euclidean distance, i.e., had found in set Qi with inquiry face picture feature descriptorRelevant all neighbours
Pairing.
6. a kind of method that recognition of face is carried out based on improvement nearest neighbor distance ratio method according to claim 1 or 5,
It is characterized in that, after executing step 4),
Neighborhood matching relevant to same feature descriptor in inquiry face picture is to being stored in the same set;
From the same incoherent neighborhood matching of feature descriptor in inquiry face picture to being stored in different set;
All set comprising all feature descriptors in inquiry face picture are stored in set P.
7. it is according to claim 1 a kind of based on the method for improving the progress recognition of face of nearest neighbor distance ratio method, it is special
Sign is, in step 6), checks the uniqueness of the included neighborhood matching pair of set P, concrete operations are as follows:
Check each key point of inquiry face picture key point whether equal in face picture to be matched corresponding thereto
Match, the matching pair of arest neighbors is retained in matched situation, completes the uniqueness inspection of set P.
8. it is according to claim 1 a kind of based on the method for improving the progress recognition of face of nearest neighbor distance ratio method, it is special
Sign is, after the uniqueness for checking the included neighborhood matching pair of set P, to institute in inquiry face picture and face picture to be matched
The matching of some key points rejects mispairing pair to using RANSAC algorithm.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910559456.4A CN110287883A (en) | 2019-06-26 | 2019-06-26 | A method of recognition of face is carried out based on nearest neighbor distance ratio method is improved |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910559456.4A CN110287883A (en) | 2019-06-26 | 2019-06-26 | A method of recognition of face is carried out based on nearest neighbor distance ratio method is improved |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110287883A true CN110287883A (en) | 2019-09-27 |
Family
ID=68005863
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910559456.4A Pending CN110287883A (en) | 2019-06-26 | 2019-06-26 | A method of recognition of face is carried out based on nearest neighbor distance ratio method is improved |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110287883A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101510257A (en) * | 2009-03-31 | 2009-08-19 | 华为技术有限公司 | Human face similarity degree matching method and device |
CN104252614A (en) * | 2013-06-27 | 2014-12-31 | 南京理工大学常熟研究院有限公司 | SIFT algorithm-based two-generation identity card face comparison method |
CN106445939A (en) * | 2015-08-06 | 2017-02-22 | 阿里巴巴集团控股有限公司 | Image retrieval, image information acquisition and image identification methods and apparatuses, and image identification system |
CN107845106A (en) * | 2017-11-17 | 2018-03-27 | 齐鲁工业大学 | Utilize the medical image registration method of improved NNDR strategies |
-
2019
- 2019-06-26 CN CN201910559456.4A patent/CN110287883A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101510257A (en) * | 2009-03-31 | 2009-08-19 | 华为技术有限公司 | Human face similarity degree matching method and device |
CN104252614A (en) * | 2013-06-27 | 2014-12-31 | 南京理工大学常熟研究院有限公司 | SIFT algorithm-based two-generation identity card face comparison method |
CN106445939A (en) * | 2015-08-06 | 2017-02-22 | 阿里巴巴集团控股有限公司 | Image retrieval, image information acquisition and image identification methods and apparatuses, and image identification system |
CN107845106A (en) * | 2017-11-17 | 2018-03-27 | 齐鲁工业大学 | Utilize the medical image registration method of improved NNDR strategies |
Non-Patent Citations (1)
Title |
---|
戚世贵等: ""一种基于图像特征点的图像匹配算法"", 《理论与方法》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2017185963A1 (en) | Big data-based method and terminal for matching trend curve local characteristics | |
CN104464079B (en) | Multiple Currencies face amount recognition methods based on template characteristic point and topological structure thereof | |
CN100414558C (en) | Automatic fingerprint distinguishing system and method based on template learning | |
CN106355577B (en) | Rapid image matching method and system based on significant condition and global coherency | |
CN109508663A (en) | A kind of pedestrian's recognition methods again based on multi-level supervision network | |
CN107871100A (en) | The training method and device of faceform, face authentication method and device | |
CN101145196B (en) | Quick fingerprint identification method based on strange topology structure | |
CN103413145B (en) | Intra-articular irrigation method based on depth image | |
CN108416342A (en) | A kind of fingerprint identification method of combination minutiae point and filament structure | |
WO2021000829A1 (en) | Multi-dimensional identity information identification method and apparatus, computer device and storage medium | |
CN101996308A (en) | Human face identification method and system and human face model training method and system | |
CN107239792A (en) | A kind of workpiece identification method and device based on binary descriptor | |
CN110309770A (en) | A kind of vehicle discrimination method again based on the study of four-tuple loss metric | |
CN101493887A (en) | Eyebrow image segmentation method based on semi-supervision learning and Hash index | |
Lu et al. | Contour box: Rejecting object proposals without explicit closed contours | |
CN105825233A (en) | Pedestrian detection method based on random fern classifier of online learning | |
CN113076891A (en) | Human body posture prediction method and system based on improved high-resolution network | |
CN102819754A (en) | Fingerprint score fusion system and method based on Sigmoid expansion | |
CN110287883A (en) | A method of recognition of face is carried out based on nearest neighbor distance ratio method is improved | |
CN105488493B (en) | A kind of quick iris identification method | |
Li et al. | An improved adaptive threshold BRISK feature matching algorithm based on SURF | |
Yang et al. | Joint multi-type feature learning for multi-modality FKP recognition | |
Zhang et al. | Eye Location Based on Adaboost and Random Forests. | |
Chen et al. | A multi-layer contrast analysis method for texture classification based on LBP | |
Chihaoui et al. | Retinal identification system based on Optical Disc Ring extraction and new local SIFT-RUK descriptor |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190927 |