CN108764068A - A kind of image-recognizing method and device - Google Patents
A kind of image-recognizing method and device Download PDFInfo
- Publication number
- CN108764068A CN108764068A CN201810431353.5A CN201810431353A CN108764068A CN 108764068 A CN108764068 A CN 108764068A CN 201810431353 A CN201810431353 A CN 201810431353A CN 108764068 A CN108764068 A CN 108764068A
- Authority
- CN
- China
- Prior art keywords
- image
- similarity
- feature vector
- group
- feature
- 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
- 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
- G06V40/161—Detection; Localisation; Normalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/30—Scenes; Scene-specific elements in albums, collections or shared content, e.g. social network photos or video
-
- 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
Abstract
The present invention discloses a kind of image-recognizing method, including：A corresponding feature vector is generated according to the feature of each image；All feature vectors are compared in pairs, calculate the similarity of each two feature vector；The image pair that similarity is higher than setting value is screened, and by the image of reservation to being ranked up according to similarity, the invention also discloses a kind of pattern recognition devices.The present invention provides student use on-line education system when authentication and teaching process in whether be student detection function, characteristics of image is extracted in image recognition processes, and the similarity between image is determined by the distance between feature vector, the smaller image of similarity is gradually deleted in comparison procedure to be calculated to simplify, recognition efficiency is high, calculation amount is saved, recognition speed is improved, saves memory space.
Description
Technical field
The present invention relates to field of face identification.More particularly, to a kind of image-recognizing method and device.
Background technology
With the popularization of computers, the fast development of internet allows online education industrial chain to welcome spring.Online education
(also referred to as e-Learning) is to be shared with Internet technology progress course content propagation by application message science and technology and quickly learned
The method of habit.The teaching method of online education is using network as medium, and by network, students and teacher is being separated by ten thousand li
Carry out education activities；In addition, by based courseware, student can also learn whenever and wherever possible, really break time and sky
Between limitation, more convenient and effective mode of learning is provided for numerous learners, especially for being busy with one's work, when study
Between for unfixed people from workplace network distance education be mode of learning the most convenient.
The equipment such as existing on-line education system, including teacher's terminal, gateway server, student terminal, teacher's terminal carry
For the video information of course, then curriculum video information is sent to student terminal by broadband gateway device by gateway server,
Student is the lecture content that may be viewed by teacher by student terminal.However, the usually used network based on cable of existing system,
It is unfavorable for the convenience that student shares learned lesson.This rapid development epoch, how better control platform resource, how to carry
Experienced for better online education becomes a great problem to student.Most on-line education system platform at present, because not being to personally instruct class
Journey can not also know to be that student teaches on acceptance line on earth so the study situation of student can not be checked directly
Educate, so as to cause occur be not I or attend class absence phenomena such as, it is very big to compare calculation amount two-by-two by recognition of face, no
It only takes and vast resources need to be consumed.
Accordingly, it is desirable to provide a kind of image-recognizing method and device.
Invention content
The purpose of the present invention is to provide a kind of image-recognizing method and devices, and student oneself can be supervised in online education
Study hard and within different learning times fast search to the same student.
In order to achieve the above objectives, the present invention uses following technical proposals：
A kind of image-recognizing method, including：
Generate the feature vector corresponding to each image respectively according to the feature of each image；
The feature vector of all generations is compared in pairs, calculates the similarity of each two feature vector；
The similarity that is calculated is screened, if the similarity of image pair higher than being retained if setting value, and by reservation
Image according to similarity to being ranked up.
Further, the similarity for calculating each two feature vector includes：Between calculating each two feature vector
Distance is higher apart from smaller then similarity.
Further, the method further includes：Image of the similarity higher than setting value is screened to rear, by the image pair of reservation
Grouping, respectively to the image in every group to pressing sequencing of similarity.
Further, described to include to carrying out sequencing of similarity to the image in every group respectively：After screening and retaining sequence
The high preceding M image pair of similarity in every group, the M are pre-set natural number.
Further, by the preceding M image in all groupings to, again according to sequencing of similarity, K is a before retaining after merging
The high image pair of similarity, the K are products of the M with image to the group number divided.
One embodiment of the invention also discloses a kind of pattern recognition device, including：
Processing module：For generating a corresponding feature vector according to the feature of each image；
Comparison module：For comparing all feature vectors in pairs, the similarity of each two feature vector is calculated；
Screening module：It is higher than the image pair of setting value for screening similarity, and by the image of reservation to according to similarity
It is ranked up.
Further, the comparison module is additionally operable to calculate the distance between each two feature vector, apart from smaller then phase
It is higher like spending.
Further, screening module be additionally operable to the image that will retain to grouping, it is similar to pressing to the image in every group respectively
Degree sequence.
Further, the screening module includes respectively to carrying out sequencing of similarity to the image in every group：It screens and protects
Stay after sequence that the high preceding M image pair of similarity, the M are pre-set natural number in every group.
Further, the screening module be additionally operable to by the preceding M image in all groupings to after merging again according to phase
Like degree sequence, the high image pair of K similarity before retaining, the K is products of the M with image to the group number divided.
One embodiment of the present of invention also discloses a kind of computer readable storage medium, the computer-readable storage medium
Instruction is stored in matter, when the computer readable storage medium is run on computers so that on the computer executes
The method stated.
Beneficial effects of the present invention are as follows：
Technical solution of the present invention provides in authentication and teaching process of the student with on-line education system when
Whether be student detection function, characteristics of image is extracted in image recognition processes, and pass through the distance between feature vector
It determining the similarity between image, gradually deletes the smaller image of similarity in comparison procedure and calculated to simplify, recognition efficiency is high,
Calculation amount is saved, recognition speed is improved, saves memory space.
Description of the drawings
Specific embodiments of the present invention will be described in further detail below in conjunction with the accompanying drawings；
Fig. 1 is one embodiment of the invention image-recognizing method flow chart；
Fig. 2 is one embodiment of the invention pattern recognition device schematic diagram；
Fig. 3 is come the structural schematic diagram of the terminal device or server system of realizing the embodiment of the present invention.
Specific implementation mode
In order to illustrate more clearly of the present invention, the present invention is done further with reference to preferred embodiments and drawings
It is bright.Similar component is indicated with identical reference numeral in attached drawing.It will be appreciated by those skilled in the art that institute is specific below
The content of description is illustrative and be not restrictive, and should not be limited the scope of the invention with this.
As shown in Figure 1, one embodiment of the invention discloses a kind of image-recognizing method, including：
S1, a corresponding feature vector is generated according to the feature of each image.
It is an object of the invention in distance education platform, from N face pictures, rapidly find out similar a pair
Face picture.So-called a pair of face picture, just assumes that picture P and picture Q, is to belong to same person, two of same person
Facial picture.Using general human face detection tech, the method for being generally based on deep learning, or it is based on HOG
(histogram oriental gradient) feature, is found out the region of face, is indicated with a rectangle frame.Detect face
Behind region, face characteristic is sought, the face characteristic of extraction is the vector of one 128 dimension.
It should be noted that the image described in embodiment includes not only facial image, or animal or plant
Image can find out pairs of image using this method in great amount of images.
S2, all feature vectors are compared in pairs, calculates the similarity of each two feature vector.
N number of facial image is compared two-by-two, the similarity compared two-by-two is come with the distance value between two feature vectors
It weighs, relatively generates a distance value every time, need to carry out N* (N-1)/2 time comparison altogether, common property gives birth to N* (N-1)/2 distance
Value.The distance between two vectors are calculated using Euclidean distance (being also Euclidean distance)：dist(X,Y)：
In formula：N=128.
X is the facial image of some student；It is 128 dimensional vectors；
Y is the facial image of another student；It is 128 dimensional vectors；
Wherein distance value dist (X, Y) is smaller, indicates that two images are more similar；If distance value is 0, then it represents that complete phase
Seemingly, as shown in table 1：
Table 1
S3, screening similarity are higher than the image pair of setting value, and by the image of reservation to being ranked up according to similarity.
To the distance value dist (X, Y) in all tables 1, total N* (N-1)/2 distance value is screened, such as according to previous
Experience retains the image pair that distance value is less than or equal to 0.4,0.4 direct exclusion is more than for the distance value between image pair, not
Consider.In this manner it is possible to filter out greatly as a result, the result of possibility less (i.e. image similarity is low) is all abandoned
Fall, only retains possibility big (apart from small, similarity big) as a result, greatly succinct subsequent calculating and storage are empty
Between, also substantially reduce the time needed for calculating.
To the image that remains to being grouped, for example, it is original have 20000* (20000-1)/2=199990000 away from
From comparison result, after being filtered according to distance threshold in above-described embodiment, only preserve 100000 as a result, by this 100000
A result is divided into 100 groups, and every group includes 1000 results, wherein first group：The first row is to the 1000th row；Second group：1001st
It goes to the 2000th row；Third group：2001st row to the 3000th row；…；100th group：99001st row to the 100000th row.
Then respectively to 1000 images in every group to being ranked up, the present embodiment is arranged according to similarity descending, i.e.,
The more forward image of arrangement is to indicating that similarity is higher, and distance value is smaller between corresponding vector.At this time to the distance value after sequence
It is screened again, only retains several pre-set in every group as a result, for example, this screening needs similarity most in total
K=5000 high image pair, according to 100 divided before groupings, every group only retains preceding M=50 in sum 1000
The high image of similarity is to that can meet condition.1000 data in every group in this way are independent mutually, do not depend on mutually, can be parallel
It is handled；Sequence processing time greatly shortens, it is parallel after time probably only originally without parallel computation 1/
100.Also can in sequence to 1000 images in every group to carrying out similarity ascending order arrangement, i.e. arrangement image pair more rearward
Indicate that similarity is higher, after retaining in this way in every group the high image of several similarities to.
Finally, total similarity is re-started after the preceding M=50 result filtered out in 100 every groups be grouped being merged
Sequence, K=5000 obtained image pair, the as highest image pair of the similarity needed for this minor sort.
As shown in Fig. 2, one embodiment of the present of invention also discloses a kind of pattern recognition device, including：
Processing module 1：For generating a corresponding feature vector according to the feature of each image；
Comparison module 2：For comparing all feature vectors in pairs, the similarity of each two feature vector is calculated；
Screening module 3：It is higher than the image pair of setting value for screening similarity, and by the image of reservation to according to similarity
It is ranked up.
Wherein, comparison module is additionally operable to calculate the distance between each two feature vector, higher apart from smaller then similarity.
Screening module is additionally operable to the image that will retain to grouping, such as 100 groups, respectively to the image in every group to pressing sequencing of similarity.
Screening module respectively to the image in every group to carry out sequencing of similarity include：It screens and retains after sequence that similarity is high in every group
Preceding M=50 image pair.By the preceding M=50 image in all groupings to, again according to sequencing of similarity, retaining after merging
The high image pair of preceding K=5000 similarity, the K are products of the M with image to the group number 100 divided.
Below with reference to Fig. 3, it illustrates suitable for the computer with the terminal device or server for realizing the embodiment of the present application
The structural schematic diagram of system.
It, can be according to being stored in read-only memory as shown in figure 3, computer system includes central processing unit (CPU)
(ROM) program in executes various appropriate from the program that storage section is loaded into random access storage device (RAM)
Action and processing.In RAM, it is also stored with various programs and data needed for system operatio.CPU, ROM and RAM pass through total
Line is connected with each other.Input/output (I/O) interface is also connected to bus.
It is connected to I/O interfaces with lower component：Include the importation of keyboard, mouse etc.；Including such as cathode-ray tube
(CRT), the output par, c of liquid crystal display (LCD) etc. and loud speaker etc.；Storage section including hard disk etc.；And including all
Such as communications portion of the network interface card of LAN card, modem.Communications portion executes logical via the network of such as internet
Letter processing.Driver is also according to needing to be connected to I/O interfaces.Detachable media, such as disk, CD, magneto-optic disk, semiconductor are deposited
Reservoir etc. is installed as needed on a drive, in order to be mounted into as needed from the computer program read thereon
Storage section.
Particularly, in accordance with an embodiment of the present disclosure, it may be implemented as computer above with reference to the process of flow chart description
Software program.For example, embodiment of the disclosure includes a kind of computer program product comprising be tangibly embodied in machine readable
Computer program on medium, the computer program include the program code for method shown in execution flow chart.At this
In the embodiment of sample, which can be downloaded and installed by communications portion from network, and/or is situated between from detachable
Matter is mounted.
Flow chart in attached drawing and block diagram, it is illustrated that according to the system of various embodiments of the invention, method and computer journey
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
A part for a part for one module, program segment, or code of table, the module, program segment, or code includes one or more
Executable instruction for implementing the specified logical function.It should also be noted that in some implementations as replacements, institute in box
The function of mark can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are practical
On can be basically executed in parallel, they can also be executed in the opposite order sometimes, this is depended on the functions involved.Also it wants
It is noted that the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart, Ke Yiyong
The dedicated hardware based system of defined functions or operations is executed to realize, or can be referred to specialized hardware and computer
The combination of order is realized.
Being described in module involved in the embodiment of the present application can be realized by way of software, can also be by hard
The mode of part is realized.Described module can also be arranged in the processor, for example, can be described as：A kind of processor packet
Include acquisition module, determining module and recommending module.Wherein, the title of these modules is not constituted under certain conditions to the module
The restriction of itself, for example, acquisition module is also described as " being used for the module of image recognition ".
As on the other hand, present invention also provides a kind of computer readable storage medium, the computer-readable storage mediums
Matter can be computer readable storage medium included in device described in above-described embodiment；Can also be individualism, not
The computer readable storage medium being fitted into terminal.There are one the computer-readable recording medium storages or more than one
Program, described program are used for executing the side for image recognition for being described in the application by one or more than one processor
Method.
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair
The restriction of embodiments of the present invention may be used also on the basis of the above description for those of ordinary skill in the art
To make other variations or changes in different ways, all embodiments can not be exhaustive here, it is every to belong to this hair
Row of the obvious changes or variations that bright technical solution is extended out still in protection scope of the present invention.
Claims (11)
1. a kind of image-recognizing method, which is characterized in that including：
Generate the feature vector corresponding to each image respectively according to the feature of each image；
The feature vector of all generations is compared in pairs, calculates the similarity of each two feature vector；
The similarity that is calculated is screened, if the similarity of image pair higher than being retained if setting value, and by the image of reservation
To being ranked up according to similarity.
2. according to the method described in claim 1, it is characterized in that, the similarity for calculating each two feature vector includes：
The distance between each two feature vector is calculated, it is higher apart from smaller then similarity.
3. according to the method described in claim 1, it is characterized in that, the method further includes：It screens similarity and is higher than setting value
Image to rear, by the image of reservation to grouping, respectively to the image in every group to pressing sequencing of similarity.
4. according to the method described in claim 3, it is characterized in that, described respectively to the image in every group to carrying out similarity row
Sequence includes：It screens and retains after sequence that the high preceding M image pair of similarity, the M are pre-set natural number in every group.
5. according to the method described in claim 4, it is characterized in that, by the preceding M image in all groupings to after merging again
According to sequencing of similarity, the high image pair of K similarity before retaining, the K is products of the M with image to the group number divided.
6. a kind of pattern recognition device, which is characterized in that including：
Processing module：For generating a corresponding feature vector according to the feature of each image；
Comparison module：For comparing all feature vectors in pairs, the similarity of each two feature vector is calculated；
Screening module：It is higher than the image pair of setting value for screening similarity, and by the image of reservation to being carried out according to similarity
Sequence.
7. device according to claim 6, which is characterized in that the comparison module is additionally operable to calculate each two feature vector
The distance between, it is higher apart from smaller then similarity.
8. device according to claim 6, which is characterized in that the screening module is additionally operable to the image that will retain to dividing
Group, respectively to the image in every group to pressing sequencing of similarity.
9. device according to claim 8, which is characterized in that the screening module is respectively to the image in every group to carrying out
Sequencing of similarity includes：It screens and retains after sequence that the high preceding M image pair of similarity, the M are pre-set in every group
Natural number.
10. device according to claim 8, which is characterized in that the screening module is additionally operable to the preceding M in all groupings
A image is to, again according to sequencing of similarity, the high image pair of K similarity before retaining, the K is M and image to dividing after merging
Group number product.
11. a kind of computer readable storage medium, which is characterized in that instruction is stored in the computer readable storage medium,
When the computer readable storage medium is run on computers so that the computer perform claim requires any in 1-5
Method described in.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810431353.5A CN108764068A (en) | 2018-05-08 | 2018-05-08 | A kind of image-recognizing method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810431353.5A CN108764068A (en) | 2018-05-08 | 2018-05-08 | A kind of image-recognizing method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108764068A true CN108764068A (en) | 2018-11-06 |
Family
ID=64010415
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810431353.5A Pending CN108764068A (en) | 2018-05-08 | 2018-05-08 | A kind of image-recognizing method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108764068A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110766077A (en) * | 2019-10-24 | 2020-02-07 | 上海眼控科技股份有限公司 | Method, device and equipment for screening sketch in evidence chain image |
| CN111382635A (en) * | 2018-12-29 | 2020-07-07 | 杭州海康威视数字技术股份有限公司 | Commodity category identification method and device and electronic equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060059112A1 (en) * | 2004-08-25 | 2006-03-16 | Jie Cheng | Machine learning with robust estimation, bayesian classification and model stacking |
| US20060106757A1 (en) * | 1996-05-06 | 2006-05-18 | Amada Company, Limited | Search for similar sheet metal part models |
| CN101263514A (en) * | 2005-08-15 | 2008-09-10 | 三菱电机株式会社 | Mutual-rank similarity-space for navigating, visualising and clustering in image databases |
| CN101833565A (en) * | 2010-03-31 | 2010-09-15 | 南京大学 | Method for actively selecting related feedbacks of representative image |
| CN103678606A (en) * | 2013-12-12 | 2014-03-26 | 天津智迅时代科技有限公司 | Method for retrieving images based on distance measurement |
-
2018
- 2018-05-08 CN CN201810431353.5A patent/CN108764068A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060106757A1 (en) * | 1996-05-06 | 2006-05-18 | Amada Company, Limited | Search for similar sheet metal part models |
| US20060059112A1 (en) * | 2004-08-25 | 2006-03-16 | Jie Cheng | Machine learning with robust estimation, bayesian classification and model stacking |
| CN101263514A (en) * | 2005-08-15 | 2008-09-10 | 三菱电机株式会社 | Mutual-rank similarity-space for navigating, visualising and clustering in image databases |
| CN101833565A (en) * | 2010-03-31 | 2010-09-15 | 南京大学 | Method for actively selecting related feedbacks of representative image |
| CN103678606A (en) * | 2013-12-12 | 2014-03-26 | 天津智迅时代科技有限公司 | Method for retrieving images based on distance measurement |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111382635A (en) * | 2018-12-29 | 2020-07-07 | 杭州海康威视数字技术股份有限公司 | Commodity category identification method and device and electronic equipment |
| CN111382635B (en) * | 2018-12-29 | 2023-10-13 | 杭州海康威视数字技术股份有限公司 | Commodity category identification method and device and electronic equipment |
| CN110766077A (en) * | 2019-10-24 | 2020-02-07 | 上海眼控科技股份有限公司 | Method, device and equipment for screening sketch in evidence chain image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Learn from all: Erasing attention consistency for noisy label facial expression recognition | |
| Akhtar et al. | Defense against universal adversarial perturbations | |
| Che et al. | How is gaze influenced by image transformations? dataset and model | |
| CN111563502B (en) | Image text recognition method and device, electronic equipment and computer storage medium | |
| CN108229651A (en) | Neural network model moving method and system, electronic equipment, program and medium | |
| CN109815955A (en) | Topic householder method and system | |
| CN113139628B (en) | Sample image identification method, device and equipment and readable storage medium | |
| CN111626126A (en) | Face emotion recognition method, device, medium and electronic equipment | |
| US20220343683A1 (en) | Expression Recognition Method and Apparatus, Computer Device, and Readable Storage Medium | |
| Zhang et al. | Research on mine vehicle tracking and detection technology based on YOLOv5 | |
| Lakkaraju et al. | A bayesian framework for modeling human evaluations | |
| CN111680678A (en) | Target area identification method, device, equipment and readable storage medium | |
| CN106227836B (en) | Unsupervised joint visual concept learning system and unsupervised joint visual concept learning method based on images and characters | |
| Zhang et al. | Web-supervised network for fine-grained visual classification | |
| CN114332911A (en) | Head posture detection method and device and computer equipment | |
| CN108764068A (en) | A kind of image-recognizing method and device | |
| Sreemathy et al. | Sign language recognition using artificial intelligence | |
| CN104966109A (en) | Medical laboratory report image classification method and apparatus | |
| JP7293658B2 (en) | Information processing device, information processing method and program | |
| KR20210097204A (en) | Methods and devices for outputting information | |
| CN113065069B (en) | Bidirectional employment recommendation method and device based on data portrait | |
| CN113516735A (en) | Image processing method, image processing device, computer readable medium and electronic equipment | |
| CN112749691A (en) | Image processing method and related equipment | |
| CN111723254A (en) | Information processing apparatus, recording medium, and information processing method | |
| Kurniawan et al. | Image processing: capturing student attendance data |
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: 20181106 |