CN110188646A - The human ear identification method merged based on gradient orientation histogram with local binary patterns - Google Patents

The human ear identification method merged based on gradient orientation histogram with local binary patterns Download PDF

Info

Publication number
CN110188646A
CN110188646A CN201910433620.7A CN201910433620A CN110188646A CN 110188646 A CN110188646 A CN 110188646A CN 201910433620 A CN201910433620 A CN 201910433620A CN 110188646 A CN110188646 A CN 110188646A
Authority
CN
China
Prior art keywords
ear
value
image
feature
pixel
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201910433620.7A
Other languages
Chinese (zh)
Other versions
CN110188646B (en
Inventor
赵立昌
陈志�
岳文静
吴宇晨
孙斗南
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
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 CN201910433620.7A priority Critical patent/CN110188646B/en
Publication of CN110188646A publication Critical patent/CN110188646A/en
Application granted granted Critical
Publication of CN110188646B publication Critical patent/CN110188646B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal 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
    • 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
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • 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
    • G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
    • G06V10/507—Summing image-intensity values; Histogram projection analysis
    • 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
    • G06V40/171—Local features and components; Facial parts ; Occluding parts, e.g. glasses; Geometrical 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
    • G06V40/172—Classification, e.g. identification
    • 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
    • G06V10/467—Encoded features or binary features, e.g. local binary patterns [LBP]

Abstract

The present invention discloses a kind of human ear identification method merged based on gradient orientation histogram with local binary patterns, solves the problems, such as that progress ear recognition discrimination is not high from ear image.The present invention extracts the gradient orientation histogram of image first, and carries out dimensionality reduction with Principal Component Analysis, then the textural characteristics of image are extracted with local binary patterns, then by two Fusion Features, is finally classified using minimum distance classifier.The present invention improves the discrimination of ear recognition by multiple features fusion, has good implementation and actual effect.

Description

The human ear identification method merged based on gradient orientation histogram with local binary patterns
Technical field
The present invention relates to a kind of human ear identification methods merged based on gradient orientation histogram with local binary patterns, belong to The interleaving techniques such as living things feature recognition, deep learning, artificial intelligence field.
Background technique
Ear recognition received state at nearly 2 years as a kind of new biometrics identification technology, theory and application research Inside and outside scholar more pays close attention to, and has important theory significance and practical application value.
Ear recognition is progress feature identification using ear image as research object, can not only be used for other biological identification technology Useful supplement, can also be applied individually to any individual identity identification occasion.In the identity recognizing technology based on biological characteristic, Ear recognition has many merits, and if ear image size is small, calculation amount is small, the distribution of color of external ear is consistent, is being converted to ash Information is lost few when spending image, and human ear is not influenced by expression shape change, and non-intrusive formula identification etc. may be implemented.
The method of ear recognition at present is divided according to extracted feature to be summarized as two major classes: one kind is based on several The method such methods of what feature construct geometrical characteristic by the key point of searching human ear profile and internal structure.Such methods It is influenced vulnerable to illumination, imaging angle, robustness is poor.One kind is the method based on algebraic characteristic, such as principle component analysis, constant III method of square, small wave converting method etc..These methods are human ear attitudes vibration is little, achieves in the preferable situation of picture quality Satisfied result.However when human ear has rotation angle change, two dimensional image can bring biggish deformation, at this moment conventional method Discrimination can sharply decline, therefore more cost effective and more accurate human ear identification method, it is also necessary to largely be ground Study carefully work.
Summary of the invention
Technical problem: the technical problem to be solved by the present invention is to how use minimum range point to the ear image of input Class device carries out ear recognition, to improve training speed and the accuracy to ear recognition.
A kind of technical solution: ear recognition side merged based on gradient orientation histogram with local binary patterns of the invention Method, comprising the following steps:
Step 1) obtains the image of human ear from ear image library；
Step 2) calculates the characteristic value in ear image in each pixel, obtains the ladder of ear image after piecemeal and standardization Spend direction histogram feature；
Step 3) carries out spatial alternation with gradient orientation histogram of the Principal Component Analysis to ear image, makes original seat Mark project to a new dimension it is lower and it is mutually orthogonal spatially, realize to the gradient orientation histogram of ear image The dimensionality reduction of feature；
Step 4) calculates the local binary patterns value of each pixel on ear image, obtains the local binary mould of ear image Formula feature；
Step 5) cascades the feature vector of gradient orientation histogram feature and local binary patterns feature, obtains new Feature vector, realize Fusion Features；
Step 6) is input to minimum distance classifier and is classified, identified.
Wherein,
The step 2) is specific as follows:
Step 21) carries out color normalization processing to the ear image obtained in step 1), uniformly converts images into ash Spend image, conversion formula are as follows: H (x, y)=0.3*R (x, y)+0.59*G (x, y)+0.11*B (x, y), wherein R (x, y), G (x, y), B (x, y) are the color-values of the red, green, blue of each pixel in image respectively, and H (x, y) indicates the ash of each pixel Angle value.
Step 22) calculates the modulus value and deflection of each pixel gradient of image using Sobel operator:
Wherein, G (x, y) indicates that the gradient magnitude of pixel, α (x, y) indicate that the gradient direction of pixel, H (x, y) indicate The gray value of pixel.
Step 23) space and direction cell weighted voting.First calculate directional weighting: x (i)=cos (θ), y (i)= Sin (θ), θ=θ+π/(Ndirection+ 1), in which: i is direction serial number；θ is angle, initial value 0；X (i) is x-axis difference in the side i To weight；Y (i) is weight of the y-axis difference in the direction i, NdirectionFor total direction number, it is traditionally arranged to be 9；
Step 24) calculates amplitude and direction, amplitude are the mean-square values of the image difference of x or y-axis, and direction value is taken as each side Upward cum rights maximum value；
Step 25) structure block simultaneously standardizes, and the feature in cell is summarized and combines blocking, calculation method are as follows: Wherein it is total to respectively indicate block x-axis by B (x), B (y) Numerical value and y-axis total value, C (x), C (y) respectively represent cell x-axis total value and y-axis total value, and B (size) is the big of block Small, B (step) is the step-length of block variation；
Step 26) summarizes the characteristic value on different directions and block, constructs the histograms of oriented gradients of image.
The step 3) is specific as follows:
Step 31) calculates the mean value of the gradient orientation histogram feature of corresponding pixel points in all ear images
Step 32) basisCovariance matrix is calculated, wherein xiFor the spy for needing dimensionality reduction Sign, UTFor covariance matrix.
Step 33) takes p principal component before covariance matrix, to each gradient orientation histogram characteristic value in ear image Feature Dimension Reduction is carried out, the gradient orientation histogram feature of the Principal Component Analysis dimensionality reduction of ear image, vector dimension p are obtained Dimension.Feature dimension reduction method isWherein y indicates principal component feature.P passes through experiment according to the actual situation and obtains, Excessive speeds can be slow, too small to will affect accuracy rate.
The step 4) is specific as follows:
The center pixel that step 41) defines 3 × 3 windows is threshold value, by the gray value of remaining 8 pixels successively with threshold value ratio Compared with the label greater than central pixel point is otherwise to be labeled as 0, and 8 bits of composition are the local binary mould of window thus The binary representation of formula value.Wherein (xc, yc) represent the center elements of 3 × 3 neighborhoods Element, its pixel value are ic, ipIndicating the value of other pixels in neighborhood, s (x) is sign function, it is 1 as x >=0, is otherwise 0, LBP(xc, yc) be center pixel local binary patterns value binary representation.
Step 42) is by the LBP (x of each pixelc, yc) be converted into decimal number and obtain final local binary patterns value, it converges The General Logistics Department obtains the local binary patterns feature of ear image.
The step 6) is specific as follows:
Ear image to be identified and the ear image classified known to several are obtained after step 2 to 5 processing respectively Feature vector input minimum distance classifier in, calculate separately ear image to be identified feature vector and each known classification The distance between the feature vector of ear image, the corresponding classification of minimum range is the classification of ear image to be identified.Away from It is as follows from calculation formula:
Wherein d (i) is ear image to be identified and i-th of known the distance between ear image classified, t1It (i) is the The first eigenvector of the i known ear images classified, p1For the first eigenvector of ear image to be identified, tnIt (i) is the N-th of feature vector of the i known images classified, pnFor n-th of feature vector of images to be recognized.
The utility model has the advantages that the invention adopts the above technical scheme compared with prior art, have following technical effect that
Present invention uses Principal Component Analysis to carry out dimensionality reduction to gradient orientation histogram feature, has filtered out a large amount of superfluous Remaining information, substantially increases the accuracy of ear recognition, while reducing the dimension of feature vector, improves the speed of ear recognition Degree；The present invention is merged using local binary patterns feature and gradient orientation histogram feature, overcomes the dry of some noises Disturb, day the high robustness of feature vector, the stability of ear recognition algorithm is improved, compared to using single features to be known Not, discrimination of the present invention is higher；The present invention uses minimum distance classifier, and computation complexity is lower, and speed is faster.Pass through these The application of method, improves the Stability and veracity of ear recognition, while reducing computation complexity, has system higher Cost-effectiveness, specifically:
(1) present invention carries out Classification and Identification using multiple features fusion, compares single features, has higher accuracy.
(2) present invention uses Principal Component Analysis carries out dimensionality reduction to gradient orientation histogram feature, has filtered out a large amount of Redundancy, substantially increase the accuracy of ear recognition.
(3) present invention uses local binary patterns feature and gradient orientation histogram feature, overcomes the dry of some noises Disturb, day the high robustness of feature vector, improve the stability of ear recognition algorithm.
(4) present invention uses Principal Component Analysis carries out dimensionality reduction to gradient orientation histogram feature, compared to traditional ladder Direction histogram feature is spent, the dimension of feature vector is reduced, improves the speed of entire ear recognition.
(5) minimum distance classifier based on Euclidean distance that the present invention uses compares other classifiers, computation complexity It is lower, improve the speed of ear recognition.
Detailed description of the invention
Fig. 1 is the Human bodys' response method flow based on convolutional neural networks.
Specific embodiment
In specific implementation, Fig. 1 is the Human bodys' response method flow based on convolutional neural networks.
This example is experimental subjects, the ear image comprising 77 people using University of Science & Technology, Beijing's human ear experiment library.Human ear library In each human ear have four ear images, be respectively: under normal circumstances the direct picture of human ear, human ear+30 degree and -30 degree rotation Image, direct picture of human ear under the conditions of illumination is dimmed.
In specific implementation, everyone has 4 width ear images, wherein 3 width images, for training, 1 width image is for testing.
Firstly, everyone 3 width ear images are input to system, carry out color normalization processing, using H (x, y)= 0.3* R (x, y)+0.59*G (x, y)+0.11*B (x, y) converts images into gray level image, wherein R (x, y), G (x, y), B (x, y) is the color-values of the red, green, blue of each pixel in image respectively, and H (x, y) indicates the gray value of each pixel；Make The modulus value and deflection of each pixel gradient of image are calculated with Sobel operator；Calculating directional weighting, amplitude and direction, amplitude is x Or the mean-square value of the image difference of y-axis, direction value are taken as the cum rights maximum value in all directions；Structure block simultaneously standardizes, will be single Feature in first lattice, which summarizes, combines blocking, calculation method are as follows: Wherein B (x), B (y) respectively indicate block x-axis total value and y-axis total value, C (x), C (y) respectively represent cell x-axis total value and y-axis total value, and B (size) is the size of block, and B (step) is block variation Step-length；Characteristic value on different directions and block is summarized, the histograms of oriented gradients of image is constructed.
Then, spatial alternation is carried out with gradient orientation histogram of the Principal Component Analysis to ear image, makes original seat Mark project to a new dimension it is lower and it is mutually orthogonal spatially, realize to the gradient orientation histogram of ear image The dimensionality reduction of feature.Specific method is basisCovariance matrix is calculated, wherein xiTo need dimensionality reduction Feature,For the mean value of human ear image gradient direction histogram feature, UTFor covariance matrix.Take p before covariance matrix a Principal component uses each gradient orientation histogram characteristic value in ear imageFeature Dimension Reduction is carried out, is obtained To the gradient orientation histogram feature of the Principal Component Analysis dimensionality reduction of ear image, wherein vector dimension is p dimension, y indicate it is main at Dtex sign.
Then, the center pixel for defining 3 × 3 windows is threshold value, by the gray value of remaining 8 pixels successively with threshold value ratio Compared with the label greater than central pixel point is otherwise to be labeled as 0, and 8 bits of composition are the local binary mould of window thus The binary representation of formula value.Wherein (xc, yc) represent the center elements of 3 × 3 neighborhoods Element, its pixel value are ic, ipIndicating the value of other pixels in neighborhood, s (x) is sign function, it is 1 as x >=0, is otherwise 0, LBP(xc, yc) be center pixel local binary patterns value binary representation, then convert thereof into decimal number and obtain finally Local binary patterns value, obtain the local binary patterns feature of ear image.
Finally, the feature vector of gradient orientation histogram feature and local binary patterns feature is cascaded, obtain new Feature vector, realize Fusion Features, be entered into minimum distance classifier and classified, identified.
Method particularly includes: the ear image classified known to ear image to be identified and several is passed through into above-mentioned steps respectively In the feature vector input minimum distance classifier obtained after processing, the feature vector of ear image to be identified and each is calculated separately A known the distance between feature vector of ear image classified, the corresponding classification of minimum range is ear image to be identified Classification.Distance calculation formula is as follows:
Wherein d (i) is ear image to be identified and i-th of known the distance between ear image classified, t1It (i) is the The first eigenvector of the i known ear images classified, p1For the first eigenvector of ear image to be identified, tnIt (i) is the N-th of feature vector of the i known images classified, pnFor n-th of feature vector of images to be recognized.
The above is only a preferred embodiment of the present invention, it should be pointed out that: for the ordinary skill people of the art For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered It is considered as protection scope of the present invention.

Claims (4)

1. a kind of human ear identification method merged based on gradient orientation histogram with local binary patterns, comprising the following steps:
Step 1) acquires several known ear images classified；
Step 2) calculates the characteristic value in ear image in each pixel, obtains the gradient side of ear image after piecemeal and standardization To histogram feature；
Step 3) carries out spatial alternation with gradient orientation histogram of the Principal Component Analysis to ear image, throws original coordinate Shadow it is lower to a new dimension and it is mutually orthogonal spatially, realize to the gradient orientation histogram feature of ear image Dimensionality reduction；
Step 4) calculates the local binary patterns value of each pixel on ear image, and the local binary patterns for obtaining ear image are special Sign；
Step 5) cascades the feature of gradient orientation histogram feature and local binary patterns feature after dimensionality reduction, obtains new Feature vector, realize Fusion Features；
Step 6) obtains the ear image classified known to ear image to be identified and several after step 2 to 5 processing respectively To feature vector input minimum distance classifier in, calculate the feature vector and each known classification of ear image to be identified The distance between feature vector of ear image, the corresponding classification of minimum range are the classification of ear image to be identified.
2. a kind of ear recognition side merged based on gradient orientation histogram with local binary patterns according to claim 1 Method, which is characterized in that the step 2) is specific as follows:
Step 21) carries out color normalization processing to the ear image obtained in step 1), uniformly converts images into grayscale image Picture, conversion formula are as follows: H (x, y)=0.3*R (x, y)+0.59*G (x, y)+0.11*B (x, y), wherein R (x, y), G (x, y), B (x, y) is the color-values of the red, green, blue of each pixel in image respectively, and H (x, y) indicates the gray value of each pixel；
Step 22) calculates the modulus value and deflection of each pixel gradient of image using Sobel operator:
Wherein, G (x, y) indicates that the gradient magnitude of pixel, α (x, y) indicate that the gradient direction of pixel, H (x, y) indicate pixel The gray value of point；
Step 23) space and direction cell weighted voting；Directional weighting: x (i)=cos (θ), y (i)=sin is calculated first (θ), θ=θ+π/(Ndirection+ 1), in which: i is direction serial number；θ is angle, initial value 0；X (i) is x-axis difference in the direction i Weight；Y (i) is weight of the y-axis difference in the direction i, NdirectionFor total direction number；
Step 24) calculates amplitude and direction, amplitude are the mean-square values of the image difference of x or y-axis, and direction value is taken as in all directions Cum rights maximum value；
Step 25) structure block simultaneously standardizes, and the feature in cell is summarized and combines blocking, calculation method are as follows: Wherein it is total to respectively indicate block x-axis by B (x), B (y) Numerical value and y-axis total value, C (x), C (y) respectively represent cell x-axis total value and y-axis total value, and B (size) is the big of block Small, B (step) is the step-length of block variation；
Step 26) summarizes the characteristic value on different directions and block, constructs the histograms of oriented gradients of image.
3. a kind of ear recognition side merged based on gradient orientation histogram with local binary patterns according to claim 1 Method, which is characterized in that the step 3) is specific as follows:
Step 31) calculates the mean value of the gradient orientation histogram feature of corresponding pixel points in all ear images
Step 32) basisCovariance matrix is calculated, wherein xiFor the feature for needing dimensionality reduction,For The mean value of ear image gradient orientation histogram feature, UTFor covariance matrix；
Step 33) takes p principal component before covariance matrix, carries out to each gradient orientation histogram characteristic value in ear image Feature Dimension Reduction, obtains the gradient orientation histogram feature of the Principal Component Analysis dimensionality reduction of ear image, and vector dimension is p dimension；It is special Levying dimension reduction method isWherein y indicates principal component feature.
4. a kind of ear recognition side merged based on gradient orientation histogram with local binary patterns according to claim 1 Method, which is characterized in that the step 4) is specific as follows:
The center pixel that step 41) defines 3 × 3 windows is threshold value, by the gray value of remaining 8 pixels successively with threshold value comparison, Label greater than central pixel point is otherwise to be labeled as 0, and 8 bits of composition are the local binary patterns value of window thus Binary representation；Wherein (xc,yc) central elements of 3 × 3 neighborhoods is represented, Its pixel value is ic, ipIndicate the value of other pixels in neighborhood, s (x) is sign function, is 1 as x >=0, is otherwise 0, LBP (xc,yc) be center pixel local binary patterns value binary representation；
Step 42) is by the LBP (x of each pixelc,yc) be converted into decimal number and obtain final local binary patterns value, after summarizing Obtain the local binary patterns feature of ear image.
CN201910433620.7A 2019-05-23 2019-05-23 Human ear identification method based on fusion of gradient direction histogram and local binary pattern Active CN110188646B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910433620.7A CN110188646B (en) 2019-05-23 2019-05-23 Human ear identification method based on fusion of gradient direction histogram and local binary pattern

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910433620.7A CN110188646B (en) 2019-05-23 2019-05-23 Human ear identification method based on fusion of gradient direction histogram and local binary pattern

Publications (2)

Publication Number Publication Date
CN110188646A true CN110188646A (en) 2019-08-30
CN110188646B CN110188646B (en) 2022-08-02

Family

ID=67717439

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910433620.7A Active CN110188646B (en) 2019-05-23 2019-05-23 Human ear identification method based on fusion of gradient direction histogram and local binary pattern

Country Status (1)

Country Link
CN (1) CN110188646B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111461147A (en) * 2020-04-30 2020-07-28 柳州智视科技有限公司 Binary coding organization algorithm based on image characteristics
CN111967531A (en) * 2020-08-28 2020-11-20 南京邮电大学 High-precision indoor image positioning method based on multi-feature fusion
CN115547475A (en) * 2022-12-05 2022-12-30 医修技术服务(北京)有限公司 Intelligent automatic control method based on weighing type medical consumable management cabinet

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106599870A (en) * 2016-12-22 2017-04-26 山东大学 Face recognition method based on adaptive weighting and local characteristic fusion
CN107066958A (en) * 2017-03-29 2017-08-18 南京邮电大学 A kind of face identification method based on HOG features and SVM multi-categorizers
CN107578007A (en) * 2017-09-01 2018-01-12 杭州电子科技大学 A kind of deep learning face identification method based on multi-feature fusion
CN108549868A (en) * 2018-04-12 2018-09-18 中国矿业大学 A kind of pedestrian detection method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106599870A (en) * 2016-12-22 2017-04-26 山东大学 Face recognition method based on adaptive weighting and local characteristic fusion
CN107066958A (en) * 2017-03-29 2017-08-18 南京邮电大学 A kind of face identification method based on HOG features and SVM multi-categorizers
CN107578007A (en) * 2017-09-01 2018-01-12 杭州电子科技大学 A kind of deep learning face identification method based on multi-feature fusion
CN108549868A (en) * 2018-04-12 2018-09-18 中国矿业大学 A kind of pedestrian detection method

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111461147A (en) * 2020-04-30 2020-07-28 柳州智视科技有限公司 Binary coding organization algorithm based on image characteristics
CN111967531A (en) * 2020-08-28 2020-11-20 南京邮电大学 High-precision indoor image positioning method based on multi-feature fusion
CN111967531B (en) * 2020-08-28 2022-09-16 南京邮电大学 High-precision indoor image positioning method based on multi-feature fusion
CN115547475A (en) * 2022-12-05 2022-12-30 医修技术服务(北京)有限公司 Intelligent automatic control method based on weighing type medical consumable management cabinet

Also Published As

Publication number Publication date
CN110188646B (en) 2022-08-02

Similar Documents

Publication Publication Date Title
CN106682598B (en) Multi-pose face feature point detection method based on cascade regression
CN103605972B (en) Non-restricted environment face verification method based on block depth neural network
CN109902590A (en) Pedestrian's recognition methods again of depth multiple view characteristic distance study
CN110188646A (en) The human ear identification method merged based on gradient orientation histogram with local binary patterns
CN104504383B (en) A kind of method for detecting human face based on the colour of skin and Adaboost algorithm
CN111126240B (en) Three-channel feature fusion face recognition method
CN102332086A (en) Facial identification method based on dual threshold local binary pattern
CN106971158B (en) A kind of pedestrian detection method based on CoLBP symbiosis feature Yu GSS feature
CN106096517A (en) A kind of face identification method based on low-rank matrix Yu eigenface
CN108108760A (en) A kind of fast human face recognition
CN105138951B (en) Human face portrait-photo array the method represented based on graph model
CN110991389A (en) Matching method for judging appearance of target pedestrian in non-overlapping camera view angle
CN106960185B (en) The Pose-varied face recognition method of linear discriminant deepness belief network
CN110516525A (en) SAR image target recognition method based on GAN and SVM
Guo et al. Real-time hand detection based on multi-stage HOG-SVM classifier
CN111832405A (en) Face recognition method based on HOG and depth residual error network
Satpathy et al. Extended histogram of gradients feature for human detection
CN108875668A (en) A kind of pedestrian detection method based on multiple features and cascade classifier
Madan et al. Traffic Sign Classification using Hybrid HOG-SURF Features and Convolutional Neural Networks.
CN109002770B (en) Face recognition method under low-resolution condition
CN106599778A (en) Rapid accurate human face detection method
CN103942572A (en) Method and device for extracting facial expression features based on bidirectional compressed data space dimension reduction
CN108090460B (en) Weber multidirectional descriptor-based facial expression recognition feature extraction method
CN103942545A (en) Method and device for identifying faces based on bidirectional compressed data space dimension reduction
CN109919041A (en) A kind of face identification method based on intelligent robot

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant