CN110287823A - Based on the face identification method for improving LBP operator and support vector cassification - Google Patents

Based on the face identification method for improving LBP operator and support vector cassification Download PDF

Info

Publication number
CN110287823A
CN110287823A CN201910495473.6A CN201910495473A CN110287823A CN 110287823 A CN110287823 A CN 110287823A CN 201910495473 A CN201910495473 A CN 201910495473A CN 110287823 A CN110287823 A CN 110287823A
Authority
CN
China
Prior art keywords
facial image
pixel
support vector
feature
lbp operator
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
Application number
CN201910495473.6A
Other languages
Chinese (zh)
Inventor
张田雨
赵君喜
李梅
彭剑
陈芸
吴婧漪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Post and Telecommunication University
Nanjing University of Posts and Telecommunications
Original Assignee
Nanjing Post and Telecommunication University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing Post and Telecommunication University filed Critical Nanjing Post and Telecommunication University
Priority to CN201910495473.6A priority Critical patent/CN110287823A/en
Publication of CN110287823A publication Critical patent/CN110287823A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2134—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on separation criteria, e.g. independent component analysis
    • 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
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • 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
    • G06V40/165—Detection; Localisation; Normalisation using facial parts and geometric relationships
    • 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)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Human Computer Interaction (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Geometry (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The present invention provides a kind of based on the face identification method for improving LBP operator and support vector cassification, comprising: constructs itself face database；Facial image pretreatment, detection, correction, gray proces and noise reduction including facial image；Using improved LBP operator extraction facial image feature, and principal component analytical method is combined to reduce feature vector dimension；Classified using feature vector of the support vector machines to extraction, and completes to identify.The present invention will improve LBP operator and the classification of support vector machines two combines, and by itself superiority for improving LBP Operators Algorithm, can obviously indicate characteristics of image, simply and rapidly complete recognition of face；In addition, the present invention has stronger robustness to illumination and expression shape change when carrying out facial image extraction, identification, recognition effect is good.

Description

Based on the face identification method for improving LBP operator and support vector cassification
Technical field
The invention belongs to technical field of image processing, and in particular to one kind is based on improvement LBP operator and support vector machines point The face identification method of class.
Background technique
In the recent period, biological identification technology development is very fast, and recognition of face persistently becomes research hotspot because of its distinctive advantage, but Face can influence to vary widely with many factors such as age, expression, posture, illumination, therefore, bring difficulty to research, The problems such as many solutions all have some limitations, for example recognition speed is excessively slow, and discrimination is low.
Research direction basic at present is to be merged using multi-method to solve single method bring limitation, for example, by using Recognition methods of the LBP operator in conjunction with other methods, but traditional LBP operator, definition region are the pros of 3 × 3 pixel sizes Shape, and the pixel value at center is defined as threshold value, other eight adjacent pixels are made comparisons with threshold value, obtain one eight Binary number, then it is converted into characteristic value of the decimal number as the region, thus one 28=256 kind shape is obtained State, recognition rate is slow and neighborhood is fixed, and when graphical rule changes, mistake often occurs in obtained characteristic value.
Summary of the invention
Aiming at the above defects or improvement requirements of the prior art, the present invention provides one kind based on improvement LBP operator and branch The face identification method for holding vector machine classification, can simply and rapidly complete recognition of face, and to illumination and expression shape change have compared with Strong robustness, recognition effect are good.
In order to achieve the above object, the invention adopts the following technical scheme:
A kind of face identification method based on improvement LBP operator and support vector cassification, includes the following steps:
Step S1 constructs itself face database；
Step S2, facial image pretreatment, detection, correction, gray proces and noise reduction including facial image；
Step S3 using improved LBP operator extraction facial image feature, and combines principal component analytical method to reduce feature Vector dimension；
Step S4 is classified using feature vector of the support vector machines to extraction, and completes to identify.
Further, the building of itself face database described in step S1 includes: selection at least 10 people, using adaptive Everyone at least intercepts 15 facial images to algorithm, and saves and constitute itself face database into a file.
Further, the pretreatment of facial image described in step S2 includes the following steps:
Step S21 detects facial image to smart self-encoding encoder network algorithm using thick, writes down the coordinate of characteristic point Position is corrected as picture centre using the triangle center that two eyes and mouth characteristic point are constituted, and cut；
Step S22 carries out gray proces to the facial image after cutting in step S21；
Step S23 carries out noise reduction process to the facial image after gray proces in step S22 using gaussian filtering.
Further, the size that facial image is cut in step S21 is 150 × 150.
Further, facial image feature extraction described in step S3 includes the following steps:
Facial image is divided into several square shaped cells regions using Matlab program by step S31, is calculated using round LBP Son makes the radius R of pixel number P and selected border circular areas variable, uses in unit area pixel mean value as threshold value, Other pixels and the threshold value are made comparisons, and the pixel greater than threshold value is assigned a value of 1, and the pixel less than threshold value is assigned a value of 0, obtains Eight-bit pixels coding, i.e. the LBP encoded radio of unit area；
Step S32 changes R value and selects k different radiuses, finds out pixel in region according to calculated for pixel values formula Variance, and the LBP characteristic value in facial image region is calculated.
Further, calculated for pixel values formula described in step S32 are as follows:
Wherein, P: the number of pixel in selected areas；R: the radius size of selected border circular areas；gp: pixel value it is big It is small；μ: the mean value size of pixel in region is acquired.
Compared with the prior art, the invention has the following beneficial effects:
Face identification method provided by the invention based on improvement LBP operator and support vector cassification will improve LBP and calculate Son and the classification of support vector machines two combine, and by itself superiority for improving LBP Operators Algorithm, can obviously indicate image spy Sign, simply and rapidly completes recognition of face；In addition, the present invention is to illumination and expression when carrying out facial image extraction, identification Variation has stronger robustness, and recognition effect is good.
Detailed description of the invention
Fig. 1 is overall flow figure of the invention；
Fig. 2 is the present invention using the schematic diagram for improving the progress feature extraction of LBP operator.
Specific embodiment
The present invention is described in further detail with specific embodiment with reference to the accompanying drawing.The embodiment is shown Example is shown in the accompanying drawings, and the specific embodiment described in following embodiments of the present invention is only used as specific reality of the invention Apply the exemplary illustration of mode, it is intended to be used to explain the present invention, and be not configured to limitation of the present invention.
A kind of face identification method based on improvement LBP operator and support vector cassification, as shown in Figure 1, including as follows Step:
Step S1 constructs itself face database.
Further, the building of itself face database described in above-mentioned steps includes: 10 people of selection or more, is used Adaboost adaptive algorithm, everyone at least intercepts 15 facial images, and saves and be used as known face into a file Database, while thus constituting itself face database.
Wherein, the interception of picture can be advantageously applied to target detection using mature Adaboost adaptive algorithm, Convenient for the detection and interception that function carries out facial image can be called directly in subsequent Mat lab program.
Step S2, facial image pretreatment, detection, correction, gray proces and noise reduction including facial image；Further Ground includes the following steps:
Step S21, using thick to smart self-encoding encoder (Coarse-to-Fine Auto-Encoder, CFAN) network algorithm Facial image is detected, writes down the coordinate position of the key feature points such as eyes, nose, mouth, and with two eyes and mouth The triangle center that bar these three characteristic points are constituted is corrected for picture centre, then every facial image after correction is cut Facial image for 150 × 150 size, the size can obtain optimal recognition effect；
Step S22, the facial image after cutting to above-mentioned steps carry out gray proces；
Step S23 carries out noise reduction process to the facial image after above-mentioned steps gray proces using gaussian filtering.
Step S3 using improved LBP operator extraction facial image feature, and combines PCA principal component analytical method to reduce Feature vector dimension.
Further, facial image feature extraction described in the step includes the following steps:
Facial image is divided into the identical square shaped cells region of several sizes using Matlab program by step S31, after adopt Rotation is made it have as shown in Fig. 2, making the radius R of pixel number P and selected border circular areas variable with circular LBP operator Turn invariance and gray scale invariance.Use in unit area pixel mean value as the region threshold, other pixels and the area Domain threshold value is made comparisons, and the pixel greater than threshold value is assigned a value of 1, and the pixel less than threshold value is assigned a value of 0, obtains eight-bit pixels volume Code, i.e. the LBP encoded radio of unit area；
Step S32 changes R value and selects k different radiuses, found out in region according to following calculated for pixel values formula The variance of pixel, different P values and the available different variance of R value, since variance is bigger, the variation in region is bigger, right The feature for describing the region just has the function of bigger, therefore, using variance as the weight of evaluation function, the region is calculated Final LBP characteristic value.
After the completion of feature extraction, the method for reusing PCA principal component analysis reduces the dimension of feature vector, helps speed up The speed of identification, the above-mentioned superiority based on the algorithm for improving LBP operator with itself, can obviously indicate characteristics of image, in addition, When carrying out facial image extraction, identification, there is stronger robustness to illumination and expression shape change.Wherein, above-mentioned pixel value meter Calculate formula are as follows:
Wherein, P: the number of pixel in selected areas；R: the radius size of selected border circular areas；gp: pixel value it is big It is small；μ: the mean value size of pixel in region is acquired.
Step S4 is classified using feature vector of the SVM support vector machines to extraction, and completes to identify.
It, can be using wherein 85% picture as training image, remaining 15% figure for the face database built Piece uses recognition methods feature extraction provided by the invention, the feature vector extracted as test image, to training image The feature that the face can be represented exercises supervision using the feature vector of all images as the input data of bis- classification method of SVM Inquiry learning and the data finished as training.When needing recognition of face, using driven by program and camera is called, by this hair It is bright that current face characteristic can be extracted based on the face identification method for improving LBP operator and support vector cassification, and and The trained comparing finished, it is quick face can be completed, be easily recognized.
In conclusion it is provided by the invention based on the face identification method for improving LBP operator and support vector cassification, it will It improves LBP operator and the classification of support vector machines two combines, it, can obvious table by itself superiority for improving LBP Operators Algorithm Show characteristics of image, simply and rapidly completes recognition of face；In addition, the present invention is to light when carrying out facial image extraction, identification There is stronger robustness according to expression shape change, can effectively solve to change face characteristic because of illumination larger to be difficult to Technical problem, recognition effect are good.
It should be noted that above-described embodiment is that illustrate the present invention rather than limit it, and Those skilled in the art can be designed alternative embodiment without departing from the scope of the appended claims.In claim In, word " comprising " does not exclude the presence of data or step not listed in the claims.

Claims (6)

1. a kind of based on the face identification method for improving LBP operator and support vector cassification, which is characterized in that including walking as follows It is rapid:
Step S1 constructs itself face database；
Step S2, facial image pretreatment, detection, correction, gray proces and noise reduction including facial image；
Step S3 using improved LBP operator extraction facial image feature, and combines principal component analytical method to reduce feature vector Dimension；
Step S4 is classified using feature vector of the support vector machines to extraction, and completes to identify.
2. according to claim 1 based on the face identification method for improving LBP operator and support vector cassification, feature Be, the building of itself face database described in step S1 include: selection at least 10 people, using adaptive algorithm everyone at least 15 facial images are intercepted, and saves and constitutes itself face database into a file.
3. according to claim 1 based on the face identification method for improving LBP operator and support vector cassification, feature It is, the pretreatment of facial image described in step S2 includes the following steps:
Step S21 detects facial image to smart self-encoding encoder network algorithm using thick, writes down the coordinate bit of characteristic point It sets, is corrected using the triangle center that two eyes and mouth characteristic point are constituted as picture centre, and cut；
Step S22 carries out gray proces to the facial image after cutting in step S21；
Step S23 carries out noise reduction process to the facial image after gray proces in step S22 using gaussian filtering.
4. according to claim 3 based on the face identification method for improving LBP operator and support vector cassification, feature It is, the size that facial image is cut in step S21 is 150 × 150.
5. according to claim 1 based on the face identification method for improving LBP operator and support vector cassification, feature It is, facial image feature extraction described in step S3 includes the following steps:
Facial image is divided into several square shaped cells regions using Matlab program, is made using round LBP operator by step S31 Pixel number P and selected border circular areas radius R it is variable, use in unit area pixel mean value as threshold value, other Pixel and the threshold value make comparisons, the pixel greater than threshold value is assigned a value of 1, and the pixel less than threshold value is assigned a value of 0, obtains eight Pixel coder, i.e. the LBP encoded radio of unit area；
Step S32 changes R value and selects k different radiuses, the side of pixel in region is found out according to calculated for pixel values formula Difference, and the LBP characteristic value in facial image region is calculated.
6. according to claim 5 based on the face identification method for improving LBP operator and support vector cassification, feature It is, calculated for pixel values formula described in step S32 are as follows:
Wherein, P: the number of pixel in selected areas；R: the radius size of selected border circular areas；gp: the size of pixel value；μ: Acquire the mean value size of pixel in region.
CN201910495473.6A 2019-06-10 2019-06-10 Based on the face identification method for improving LBP operator and support vector cassification Pending CN110287823A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910495473.6A CN110287823A (en) 2019-06-10 2019-06-10 Based on the face identification method for improving LBP operator and support vector cassification

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910495473.6A CN110287823A (en) 2019-06-10 2019-06-10 Based on the face identification method for improving LBP operator and support vector cassification

Publications (1)

Publication Number Publication Date
CN110287823A true CN110287823A (en) 2019-09-27

Family

ID=68003921

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910495473.6A Pending CN110287823A (en) 2019-06-10 2019-06-10 Based on the face identification method for improving LBP operator and support vector cassification

Country Status (1)

Country Link
CN (1) CN110287823A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110781800A (en) * 2019-10-23 2020-02-11 北京远舢智能科技有限公司 Image recognition system

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101727573A (en) * 2008-10-13 2010-06-09 汉王科技股份有限公司 Method and device for estimating crowd density in video image
CN108564042A (en) * 2018-04-17 2018-09-21 谭红春 A kind of facial expression recognition system based on hepatolenticular degeneration patient
CN109002770A (en) * 2018-06-25 2018-12-14 电子科技大学 A kind of face identification method under the conditions of low resolution

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101727573A (en) * 2008-10-13 2010-06-09 汉王科技股份有限公司 Method and device for estimating crowd density in video image
CN108564042A (en) * 2018-04-17 2018-09-21 谭红春 A kind of facial expression recognition system based on hepatolenticular degeneration patient
CN109002770A (en) * 2018-06-25 2018-12-14 电子科技大学 A kind of face identification method under the conditions of low resolution

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
唐朝晖等: "基于 LBPV 的浮选泡沫图像纹理特征提取", 《计算机应用研究》 *
文章: "人脸识别关键技术研究与应用", 《中国优秀硕士学位论文全文数据库信息科技辑》 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110781800A (en) * 2019-10-23 2020-02-11 北京远舢智能科技有限公司 Image recognition system

Similar Documents

Publication Publication Date Title
CN111339990B (en) Face recognition system and method based on dynamic update of face features
CN106815566B (en) Face retrieval method based on multitask convolutional neural network
CN111368683B (en) Face image feature extraction method and face recognition method based on modular constraint CenterFace
CN114067143B (en) Vehicle re-identification method based on double sub-networks
CN109376604B (en) Age identification method and device based on human body posture
CN109919039B (en) Static gesture recognition method based on palm and finger characteristics
CN108038434B (en) Video facial expression pre-detection method based on multi-example learning
CN108369644B (en) Method for quantitatively detecting human face raised line, intelligent terminal and storage medium
CN104408449A (en) Intelligent mobile terminal scene character processing method
CN110826408B (en) Face recognition method by regional feature extraction
CN106127193B (en) A kind of facial image recognition method
CN112597812A (en) Finger vein identification method and system based on convolutional neural network and SIFT algorithm
CN107045621A (en) Facial expression recognizing method based on LBP and LDA
CN110728302A (en) Method for identifying color textile fabric tissue based on HSV (hue, saturation, value) and Lab (Lab) color spaces
CN109086772A (en) A kind of recognition methods and system distorting adhesion character picture validation code
KR101641500B1 (en) Fast Eye Detection Method Using Block Contrast and Symmetry in Mobile Device
CN104021384A (en) Face recognition method and device
Lee et al. Robust fingertip extraction with improved skin color segmentation for finger gesture recognition in Human-robot interaction
CN113159045A (en) Verification code identification method combining image preprocessing and convolutional neural network
CN105205460A (en) Face expression feature extraction and recognition method based on maximum direction encoding
CN108764062B (en) Visual sense-based clothing piece identification method
CN104376312A (en) Face recognition method based on word bag compressed sensing feature extraction
CN110287823A (en) Based on the face identification method for improving LBP operator and support vector cassification
CN111612083B (en) Finger vein recognition method, device and equipment
CN112308044B (en) Image enhancement processing method and palm vein identification method for palm vein image

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

Application publication date: 20190927

RJ01 Rejection of invention patent application after publication