CN102073842A - Method for recognizing faces through classified matching - Google Patents
Method for recognizing faces through classified matching Download PDFInfo
- Publication number
- CN102073842A CN102073842A CN2009103103460A CN200910310346A CN102073842A CN 102073842 A CN102073842 A CN 102073842A CN 2009103103460 A CN2009103103460 A CN 2009103103460A CN 200910310346 A CN200910310346 A CN 200910310346A CN 102073842 A CN102073842 A CN 102073842A
- Authority
- CN
- China
- Prior art keywords
- face
- picture
- correlation
- angle
- database
- 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
Images
Landscapes
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a method for recognizing faces through classified matching, aiming to provide a method for recognizing faces by classification and areas before correlation algorithm so as to improve the accuracy and operation speed and adapt to rapid recognition requirement. The method comprises the following steps: establishing a face database; inputting face images, computing the direction of each point on the face, connecting the directions with smooth lines, finding out the core angle and determining the class of the core angle; determining the angles among the eyes, the mouth and the nose according to the class of the core angle and determining the class of image areas; and carrying out correlation matching between the obtained class of the relevant area and the information in the face database by a convolution method, finding out the degrees of correlation of the images in the database and selecting the image with the highest degrees of correlation as the target image. The method has the following technical effects: classifying the input face images before correlation algorithm and then carrying out correlation computation by the convolution method, thus reducing the operation time, reducing the probability of errors and improving the accuracy and the operation speed.
Description
Technical field
The present invention relates to a kind of method of recognition of face.
Background technology
Along with developing rapidly of digital camera and correlation technique, the correlation techniques of a lot of relevant recognitions of face have appearred on the market, as recognition of face, tooth identification is leaked in smiling face's identification.Because it is peculiar that the digital camera technology of these identifications and earlier generations has, and captured desired people's face field well, this technology is subjected to the consumer and welcomes widely.Face recognition technology and fingerprint recognition are similar to the ballot login in addition, driving license, and the welfare certificate, prevention is swindled and is used for the comprehensive identity validation of establishing of similar application aspect and used more and more.
Though these technology have all been used the same basic fundamental of recognition of face, because the top recognition technology all is directly to adopt relevance algorithms, and, what adopt is fast fourier transform method, the computation process complexity, and computing time is long, correctness is poor, and travelling speed is slow.With various environment and purpose like this, can not carry out recognition of face fast.
Summary of the invention
The present invention is in order to overcome weak point of the prior art, provide a kind of before relevance algorithms to people's face classify, subregion identification, to improve correctness and travelling speed, adapt to the face identification method of quick identification needs.
The present invention is achieved through the following technical solutions:
A kind of face identification method by classification and matching is characterized in that, comprises the steps:
(1) set up face database, wherein, the people's face information in people's face taxonomy database comprises: the kind at each core angle and coordinate also have the angle information between the core angle;
(2) import people's face picture, calculate the direction of facial each point, connect all directions, find out the core angle with level and smooth lines, and the kind at definite core angle; Determine angle between eyes, mouth, the nose according to the kind at core angle, determine the kind of image region;
(3) kind and the information in the face database with the relevant range that obtains adopts the method for convolution to carry out relevant matches, find out with database in the degree of correlation of picture, select the highest picture of the degree of correlation as Target Photo.
People's face picture of input is the jpg picture in the step (2), treats the color and the size criteriaization of the picture of input before the input picture earlier.
Definite method at core angle comprises the steps: in the step (2)
(1) each pixel in the whole picture is converted into value between 0～1;
(2) pixel with per 9 (3x3) is a unit, according to the descending direction of deriving set of pixels integral body of value;
(3) according to the much directions of such set of pixels that have direction in the picture, the trend of the set of pixels direction on the picture that draws;
(4) walk upwards from this, which can determine is core point.
The present invention has following technique effect:
Face identification method of the present invention is classified to people's face picture of input earlier before carrying out relevance algorithms, adopt convolution method to carry out correlation calculations again, significantly reduced working time, and reduced the probability of makeing mistakes, and improved correctness and travelling speed, be applicable to the occasion of quick identification, give people with the further sense of security in life, and, when getting driving license, can not waste the too many time such as the ballot login.
Description of drawings
Fig. 1 does not carry out the more coarse embodiment that 0～1 value transforms for the present invention;
Fig. 2 is the computing method diagram;
Angle determines between Fig. 3 eye, nose, the mouth;
Fig. 4 is the process flow diagram of sorting algorithm of the present invention;
Fig. 5 is the process flow diagram of relevant matches algorithm of the present invention;
Fig. 6 is a generic point;
Fig. 7 is a trigpoint;
Fig. 8 is a core point.
Embodiment
If the RGB picture is converted to digital form, then dim spot value is 0, and bright spot value is 255, between bright-dark degree all be that precentagewise is calculated.Again they are converted into the value between 0～1.
Fig. 1 is more coarse example, it is not carried out 0～1 value and transforms, and the picked-up of actual point is much closeer than it.Wherein we to get at 9 be 1 unit, calculate the value of each point, it is big to little direction and whole 9 mean values to determine having of it roughly according to its value, is direction and the mean value that a unit calculates 9 units in the same manner with 9 points again, go down by that analogy, as shown in Figure 2.
Below in conjunction with the drawings and specific embodiments to the detailed description of the invention.
Face identification method by classification and matching of the present invention comprises the steps:
(1) set up face database, wherein, the people's face information in the face database comprises: the value of each each point set of picture and direction are the same with the top pictorial information disposal route of mentioning.For instance, have a cover system to use in a company, then need in advance the people's face information of all personnel in the company to be preserved, set up a storehouse, data among concrete people and the Ku being made comparisons during actual the use gets final product again.This method is convenient more reliable than gate control system on some degree.
(2) input people face picture at first carries out sorting algorithm, and Fig. 4 is the direction that the process flow diagram of sorting algorithm of the present invention calculates facial each point, connects all directions with level and smooth lines, finds out the core angle, and the kind at definite core angle; Determine angle between eyes, mouth, the nose according to the kind at core angle, determine the kind of image region.
(3) carry out relevant matches, Fig. 5 is the process flow diagram of relevant matches algorithm of the present invention, adopt the method for convolution to carry out relevant matches the kind and the information in the face database of the relevant range that obtains, find out with database in the degree of correlation of picture, select the highest picture of the degree of correlation as Target Photo.
Flow sequence is: picture obtains-black and white conversion-big or small homogeneous-quantification (aiming at digital form)-determine angle between the kind-core angle point of the direction of the direction of each point-determine point set-connect each point direction (smoothing)-find out and definite core angle point with level and smooth curve calculate fixed-contrast with data in the database.
Convolution method: regard new picture as the f function, pictorial information is regarded the g function as in the storehouse, carries out convolution.Can regard after upset of one of them function and the translation integration as with the product of another function.
(f＊g)(t)＝∫f(τ)g(t-τ)dτ。
Claims (3)
1. the face identification method by classification and matching is characterized in that,
Comprise the steps:
(1) set up face database, wherein, the people's face information in people's face taxonomy database comprises: the kind at each core angle and coordinate also have the angle information between the core angle;
(2) import people's face picture, calculate the direction of facial each point, connect all directions, find out the core angle with level and smooth lines, and the kind at definite core angle; Determine angle between eyes, mouth, the nose according to the kind at core angle, determine the kind of image region;
(3) kind and the information in the face database with the relevant range that obtains adopts the method for convolution to carry out relevant matches, find out with database in the degree of correlation of picture, select the highest picture of the degree of correlation as Target Photo.
2. the face identification method by classification and matching according to claim 1 is characterized in that, people's face picture of input is the jpg picture in the step (2), treats the color and the size criteriaization of the picture of input before the input picture earlier.
3. the face identification method by classification and matching according to claim 1 is characterized in that, definite method at core angle comprises the steps: in the step (2)
(1) each pixel in the whole picture is converted into value between 0～1;
(2) pixel with per 9 (3x3) is a unit, according to the descending direction of deriving set of pixels integral body of value;
(3) according to the much directions of such set of pixels that have direction in the picture, the trend of the set of pixels direction on the picture that draws;
(4) walk upwards from this, which can determine is core point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009103103460A CN102073842A (en) | 2009-11-25 | 2009-11-25 | Method for recognizing faces through classified matching |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009103103460A CN102073842A (en) | 2009-11-25 | 2009-11-25 | Method for recognizing faces through classified matching |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102073842A true CN102073842A (en) | 2011-05-25 |
Family
ID=44032376
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2009103103460A Pending CN102073842A (en) | 2009-11-25 | 2009-11-25 | Method for recognizing faces through classified matching |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102073842A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542299A (en) * | 2011-12-07 | 2012-07-04 | 惠州Tcl移动通信有限公司 | Face recognition method, device and mobile terminal capable of recognizing face |
| CN103151020A (en) * | 2013-02-18 | 2013-06-12 | 明基电通有限公司 | Display device |
| CN103324880A (en) * | 2012-03-15 | 2013-09-25 | 欧姆龙株式会社 | Authentication apparatus, and control method of authentication apparatus |
| CN108304838A (en) * | 2017-05-11 | 2018-07-20 | 腾讯科技（深圳）有限公司 | A kind of pictorial information recognition methods and terminal |
-
2009
- 2009-11-25 CN CN2009103103460A patent/CN102073842A/en active Pending
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542299A (en) * | 2011-12-07 | 2012-07-04 | 惠州Tcl移动通信有限公司 | Face recognition method, device and mobile terminal capable of recognizing face |
| CN102542299B (en) * | 2011-12-07 | 2015-03-25 | 惠州Tcl移动通信有限公司 | Face recognition method, device and mobile terminal capable of recognizing face |
| CN103324880A (en) * | 2012-03-15 | 2013-09-25 | 欧姆龙株式会社 | Authentication apparatus, and control method of authentication apparatus |
| CN103324880B (en) * | 2012-03-15 | 2016-08-10 | 欧姆龙株式会社 | Certification device and the control method of certification device |
| CN103151020A (en) * | 2013-02-18 | 2013-06-12 | 明基电通有限公司 | Display device |
| CN108304838A (en) * | 2017-05-11 | 2018-07-20 | 腾讯科技（深圳）有限公司 | A kind of pictorial information recognition methods and terminal |
| CN108304838B (en) * | 2017-05-11 | 2020-01-10 | 腾讯科技（深圳）有限公司 | Picture information identification method and terminal |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3740897B1 (en) | License plate reader using optical character recognition on plural detected regions | |
| KR102596897B1 (en) | Method of motion vector and feature vector based fake face detection and apparatus for the same | |
| WO2019169532A1 (en) | License plate recognition method and cloud system | |
| US10025998B1 (en) | Object detection using candidate object alignment | |
| Greenhalgh et al. | Recognizing text-based traffic signs | |
| CN109145742B (en) | Pedestrian identification method and system | |
| US9842266B2 (en) | Method for detecting driver cell phone usage from side-view images | |
| WO2015165365A1 (en) | Facial recognition method and system | |
| US10445602B2 (en) | Apparatus and method for recognizing traffic signs | |
| WO2018216629A1 (en) | Information processing device, information processing method, and program | |
| US9489561B2 (en) | Method and system for estimating fingerprint pose | |
| JP2018198053A (en) | Information processor, information processing method, and program | |
| US10089525B1 (en) | Differentiating left and right eye images | |
| KR101937323B1 (en) | System for generating signcription of wireless mobie communication | |
| US9710707B1 (en) | Detecting iris orientation | |
| US11210548B2 (en) | Railroad track recognition device, program, and railroad track recognition method | |
| US9846807B1 (en) | Detecting eye corners | |
| CN103218610A (en) | Formation method of dogface detector and dogface detection method | |
| CN102073842A (en) | Method for recognizing faces through classified matching | |
| US20110142345A1 (en) | Apparatus and method for recognizing image | |
| CN104021372A (en) | Face recognition method and device thereof | |
| JP2013218605A (en) | Image recognition device, image recognition method, and program | |
| JP2022538127A (en) | Slap segmentation of contactless fingerprint images | |
| Cai et al. | Robust facial expression recognition using RGB-D images and multichannel features | |
| TW201627912A (en) | Image processing system and method for license plate recognition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20110525 |