CN105678208B - Method and device for extracting face texture - Google Patents
Method and device for extracting face texture Download PDFInfo
- Publication number
- CN105678208B CN105678208B CN201510191874.4A CN201510191874A CN105678208B CN 105678208 B CN105678208 B CN 105678208B CN 201510191874 A CN201510191874 A CN 201510191874A CN 105678208 B CN105678208 B CN 105678208B
- Authority
- CN
- China
- Prior art keywords
- image
- texture
- face
- initial texture
- group
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000001914 filtration Methods 0.000 claims abstract description 70
- 230000001815 facial effect Effects 0.000 claims description 93
- 238000012876 topography Methods 0.000 claims description 63
- 210000001508 eye Anatomy 0.000 claims description 54
- 210000004709 eyebrow Anatomy 0.000 claims description 54
- 210000000056 organ Anatomy 0.000 claims description 30
- 239000000284 extract Substances 0.000 claims description 18
- 238000000605 extraction Methods 0.000 claims description 18
- 238000012935 Averaging Methods 0.000 claims description 13
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 238000012163 sequencing technique Methods 0.000 claims description 6
- 230000002349 favourable effect Effects 0.000 abstract 1
- 210000000214 mouth Anatomy 0.000 description 50
- 238000010586 diagram Methods 0.000 description 13
- 210000000887 face Anatomy 0.000 description 6
- 210000001331 nose Anatomy 0.000 description 6
- 210000004209 hair Anatomy 0.000 description 5
- 230000015556 catabolic process Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008921 facial expression Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000002156 mixing Methods 0.000 description 2
- 210000005252 bulbus oculi Anatomy 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
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
- G06V40/168—Feature extraction; Face representation
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for extracting face texture, which comprises the following steps: reading a face image; 2D-Gabor filtering processing is carried out on the face image to generate a plurality of initial texture images; grouping the plurality of initial texture images on average; and respectively extracting characteristic images from each group of initial texture images and outputting the characteristic images. The invention also discloses a device for extracting the face texture. The method eliminates some unimportant texture features in the initial texture image, so that the extracted face texture features are more prominent, and the method is favorable for recognition processing.
Description
Technical field
The present invention relates to field of image processing more particularly to a kind of method and devices for extracting face texture.
Background technique
2D-Gabor filter is a kind of variation of Fourier in short-term, can achieve the suboptimization of airspace and frequency domain.Benefit
When being extracted with 2D-Gabor filter to face texture, it is often necessary to carry out the 2D-Gabor filter of the multiple scales of multiple directions
Wave has good direction selection and frequency in spatial domain and frequency domain, and the face characteristic of extraction is more comprehensive, Ke Yiyou
The information of the one width facial image of expression of effect.It is filtered by 2D-Gabor at present during extracting face texture, due to texture
Feature divides thinner, and there are the textural characteristics of extraction are not significant, some similar textural characteristics can interfere face characteristic to know
Other problem.
Above content is only used to facilitate the understanding of the technical scheme, and is not represented and is recognized that above content is existing skill
Art.
Summary of the invention
The main purpose of the present invention is to provide a kind of methods for extracting face texture, it is intended to solve existing to pass through 2D-
During face texture is extracted in Gabor filtering, the inapparent problem of face texture information feature of extraction.
To achieve the above object, the present invention provides a kind of method for extracting face texture, the side for extracting face texture
Method includes the following steps:
Read facial image；
2D-Gabor filtering processing is carried out to the facial image, generates multiple initial texture images；
Average packet is carried out to the multiple initial texture image；
Characteristic image is extracted from every group of initial texture image respectively, and exports the characteristic image.
Preferably, described to extract characteristic image from every group of initial texture image respectively, and export the characteristic image
Step includes:
Each initial texture image averaging in each group is decomposed into image block；
The maximum image block of same position pixel value in every group is determined, as characteristic image block；
Every group of characteristic image block is combined into characteristic image, and exports the characteristic image.
Preferably, described that 2D-Gabor filtering processing is carried out to the facial image, generate multiple initial texture images
Step includes:
The topography containing face organ is extracted from the facial image；
Whole to the facial image and described topography carries out 2D-Gabor filtering processing respectively, generates multiple initial
Texture image.
Preferably, the step of topography extracted from the facial image containing face organ includes:
The image-region in the facial image containing face organ is determined according to preset ratio；
The topography containing eyes, eyebrow and/or mouth is extracted from described image region.
Preferably, described the step of carrying out average packet to the multiple initial texture image, includes:
The initial texture image of the same direction is ranked up by scale；
Two initial texture images of the adjacent scale of the same direction are divided into one group, successively carry out average packet.
In addition, to achieve the above object, the present invention also provides a kind of device for extracting face texture, the extraction face lines
The device of reason includes:
Read module, for reading facial image；
Filter module generates multiple initial texture images for carrying out 2D-Gabor filtering processing to the facial image；
Grouping module, for carrying out average packet to the multiple initial texture image；
Extraction module for extracting characteristic image from every group of initial texture image respectively, and exports the characteristic image.
Preferably, the extraction module includes decomposition unit, determination unit and assembled unit；
The decomposition unit, for each initial texture image averaging in each group to be decomposed into image block；
The determination unit, for determining the maximum image block of same position pixel value in every group, as characteristic image block；
The assembled unit for every group of characteristic image block to be combined into characteristic image, and exports the characteristic image.
Preferably, the filter module includes acquiring unit and filter unit；
The acquiring unit, for extracting the topography containing face organ from the facial image；
The filter unit, for whole to the facial image and the topography to carry out 2D-Gabor filtering respectively
Processing, generates multiple initial texture images.
Preferably, the acquiring unit is also used to determine in the facial image according to preset ratio and contains face organ
Image-region；
The acquiring unit is also used to extract the Local map containing eyes, eyebrow and/or mouth from described image region
Picture.
Preferably, the grouping module includes sequencing unit and division unit；
The sequencing unit, for the initial texture image of the same direction to be ranked up by scale；
The division unit, for two initial texture images of the adjacent scale of the same direction to be divided into one group, successively
Carry out average packet.
The present invention is grouped by the multiple initial texture image averagings for generating 2D-Gabor filtering processing, and at the beginning of each group
Characteristic image is extracted in beginning texture image, eliminates some unessential textural characteristics in initial texture image, so that extracting
Face textural characteristics it is more prominent, be conducive to identifying processing.
Detailed description of the invention
Fig. 1 is the flow diagram of the first embodiment for the method that the present invention extracts face texture；
Fig. 2 is the refinement flow diagram of step S40 in Fig. 1；
Fig. 3 is the flow diagram of the second embodiment for the method that the present invention extracts face texture；
Fig. 4 is the structural schematic diagram for the preferred embodiment that the present invention is divided facial image by preset ratio；
Fig. 5 is the process for the preferred embodiment that the present invention extracts the topography containing face organ from facial image
Schematic diagram；
Fig. 6 is the functional block diagram of the first embodiment for the device that the present invention extracts face texture；
Fig. 7 is the functional block diagram of the preferred embodiment of grouping module in Fig. 6；
Fig. 8 is the functional block diagram of the preferred embodiment of extraction module in Fig. 6；
Fig. 9 is the functional block diagram of the second embodiment for the device that the present invention extracts face texture.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
The primary solutions of the embodiment of the present invention are: reading facial image；2D-Gabor is carried out to the facial image
Filtering processing, generates multiple initial texture images；Average packet is carried out to the multiple initial texture image；Respectively at the beginning of every group
Characteristic image is extracted in beginning texture image, and exports the characteristic image.
During the existing filtering extraction face texture by 2D-Gabor, the face texture information feature of extraction is not significant
The problem of.
Based on the above issues, the present invention provides a kind of method for extracting face texture.
Referring to Fig.1, Fig. 1 is the flow diagram of the first embodiment for the method that the present invention extracts face texture.
In the present embodiment, the method for extracting face texture includes:
Step S10 reads facial image；
Step S20 carries out 2D-Gabor filtering processing to the facial image, generates multiple initial texture images；
The facial image of face textural characteristics to be extracted is read, the facial image, which can be, to carry out in face recognition process
Facial image to be identified is also possible to the sample image for comparison, or is also possible to other face textures to be extracted
Facial image.The initial texture image can be one or more than one initial texture image, it is preferred that with image
The form of matrix is indicated.The 2D-Gabor filtering processing can be one-dimensional 2D-Gabor filtering processing, be also possible to two
2D-Gabor filtering processing is tieed up, the two-dimensional filtering processing is included in N (N is positive integer) a direction M (M is positive integer) a scale
2D-Gabor filtering processing is carried out, N*M initial texture image is generated.In the present embodiment, preferably to the facial image
Carry out two dimension 2D-Gabor filtering processing.It should be understood that carrying out 2D-Gabor filtering processing to the face, however it is not limited to
2D-Gabor filtering processing is integrally carried out to the facial image, is also possible to carry out 2D- to the topography of the face
Gabor filtering processing, or to the facial image is whole and the topography of the facial image carries out 2D-Gabor respectively
Filtering processing.Such as: it is whole to the facial image to carry out two dimension 2D-Gabor filtering processing in M, N number of direction scale, it generates
N*M initial texture image；Or the topography to the facial image, topography such as comprising eyes eyebrow and comprising
The topography of mouth, M, N number of direction, scale carries out two dimension 2D-Gabor filtering processing respectively, generates 2*N*M initial lines
Manage image；Or to the facial image entirety, the topography comprising eyes eyebrow and comprising the topography of mouth, divide
The two-dimentional 2D-Gabor filtering processing for not carrying out M, N number of direction scale, generates 3*N*M initial texture image.
Step S30 carries out average packet to the multiple initial texture image；
The initial texture image of the same direction is ranked up by scale；By two initial lines of the adjacent scale of the same direction
Reason image is divided into one group, successively carries out average packet.Two initial texture images of the adjacent scale of the same direction can be drawn
It is divided into one group, average packet can also be carried out in such a way that more than two initial texture images are divided into one group.Such as:
It is whole to the facial image to carry out two dimension 2D-Gabor filtering in a scale of N (N is positive integer) a direction M (M is positive integer)
Processing generates N*M initial texture image, the initial texture image of the same direction is ranked up by scale, by the same direction
Two initial texture images of adjacent scale are divided into one group, successively carry out average packet, by first initial line of all directions
Reason image and second initial texture image are divided into one group, by the third initial texture image of all directions and the 4th initial line
Reason image be divided into one group, and so on obtain a initial texture image group of N* (M/2)；Alternatively, to the part of the facial image
Image, the topography such as comprising eyes eyebrow and the topography comprising mouth, M, N number of direction, scale carries out two respectively
2D-Gabor filtering processing is tieed up, 2*N*M initial texture image is generated, by the same direction of the initial texture image of eyes eyebrow
Two initial texture images of adjacent scale are divided into one group, successively carry out average packet, at the beginning of obtaining a eyes eyebrow of N* (M/2)
Two initial texture images of the adjacent scale of the same direction of the initial texture image of mouth are divided into one by beginning texture image group
Group successively carries out average packet, obtains a mouth initial texture image group of N* (M/2)；Alternatively, whole to the facial image,
Topography comprising eyes eyebrow and the topography comprising mouth carry out the two-dimentional 2D- of M, N number of direction scale respectively
Gabor filtering processing generates the initial texture image of N*M face entirety, the initial texture image and N* of N*M eyes eyebrow
The initial texture image of M mouth, by two initial lines of the adjacent scale of the same direction of the initial texture image of face entirety
Reason image is divided into one group, successively carries out average packet, the initial texture image group of a face entirety of N* (M/2) is obtained, by eye
Two initial texture images of the adjacent scale of the same direction of the initial texture image of eyeball eyebrow are divided into one group, successively carry out flat
It is grouped, obtains a eyes eyebrow initial texture image group of N* (M/2), the same direction of the initial texture image of mouth is adjacent
Two initial texture images of scale are divided into one group, successively carry out average packet, obtain a mouth initial texture figure of N* (M/2)
As group.
Step S40 extracts characteristic image from every group of initial texture image respectively, and exports the characteristic image.
The characteristic image that can represent face textural characteristics is extracted most from every group of initial texture image respectively, and will
The characteristic image output.Such as: to the facial image entirety, the topography comprising eyes eyebrow and include the office of mouth
Portion's image carries out the two-dimentional 2D-Gabor filtering processing of N (N is positive integer) a direction M (M is positive integer) a scale respectively, raw
At the initial line of the initial texture image of N*M face entirety, the initial texture image of N*M eyes eyebrow and N*M mouth
Image is managed, one group is divided into according to by two initial texture images of the adjacent scale of the same direction to the initial texture image
Mode carries out average packet, respectively obtains initial texture image group, (N*M)/2 eyes eyebrow of a face entirety of N* (M/2)
Initial texture image group and a mouth initial texture image group of N* (M/2), extract from every group of initial texture image respectively
The characteristic image that face textural characteristics can most be represented obtains the characteristic image of a face entirety of N* (M/2), N* (M/2) a eyes
The characteristic image of characteristic image and N* (M/2) a mouth of eyebrow.Referring to Fig. 2, can be achieved by the steps of:
Each initial texture image averaging in each group is decomposed into image block by step S401；
Initial texture image in each group can be equally divided into i (i is positive integer) part, width direction in the height direction
On be equally divided into j (j is positive integer) part, each initial texture picture breakdown obtains i*j image block, and the short transverse is institute
The vertical direction of initial texture image is stated, the width direction is the horizontal direction of the initial texture image, it is preferred that will be every
The i*j image block that a initial texture picture breakdown obtains is indicated with the image array that i row j is arranged.
Step S402 determines the maximum image block of same position pixel value in every group, as characteristic image block；
Determine the pixel value of each image block, it is preferred that determine that the process of the pixel value of each image block can be, one by one
The pixel value for inquiring each pixel of described image block, takes the pixel value of the maximum pixel of pixel value as described image block
Pixel value.By in every group, the pixel value of the image block of same position is compared, and takes the maximum image block of pixel value as feature
Image block.Such as: have the first initial texture image and the second initial texture image in an initial texture image group, by this two
A initial texture image averaging is decomposed into i*j (i is positive integer, and j is positive integer) a image block, that is, by initial texture image
Height is equally divided into i equal portions, and the width of initial texture image is equally divided into j equal portions.The pixel value for determining each image block, is looked into one by one
The pixel value for asking each pixel of described image block, takes pixel of the pixel value of the maximum point of pixel value as described image block
Value.I-th row jth of the pixel value of the i-th row jth column image block of the first initial texture image and the second initial texture image is arranged
The pixel value of image block compares, and takes the maximum image block of pixel value as characteristic image block.Every group obtains i*j characteristic pattern
As block.
Every group of characteristic image block is combined into characteristic image, and exports the characteristic image by step S403.
By the characteristic image block in every group by preset order permutation and combination at characteristic image, and export the characteristic image.
Such as: the characteristic image block extracted is arranged from i-th (i is positive integer) row jth (j is positive integer) of initial texture image, then
Come characteristic image the i-th row jth column, and so on carry out characteristic image combination, using the characteristic image as representative
The image of face textural characteristics exports.
The present embodiment is grouped by the multiple initial texture image averagings for generating 2D-Gabor filtering processing, and from each group
Characteristic image is extracted in initial texture image, some unessential textural characteristics in initial texture image are eliminated, so that mentioning
The face textural characteristics taken are more prominent, are conducive to identifying processing.
It is the flow diagram of the second embodiment for the method that the present invention extracts face texture referring to Fig. 3, Fig. 3.Based on upper
The first embodiment for extracting the method for face texture is stated, the step S20 includes:
Step S21 extracts the topography containing face organ from the facial image；
Facial image is read, the topography containing face organ, the face organ are extracted from the facial image
Including eyes, eyebrow, nose or mouth etc..It can be extracted from the facial image containing eyes, eyebrow, nose or mouth
The topography of equal face organs' one of which is handled；It can also be extracted from the facial image containing eyes, eyebrow
More than one the topography of the face organs such as hair, mouth or nose is handled.Preferably, it is mentioned from the facial image
Take out the topography containing eyes eyebrow, mouth, during carrying out facial expression recognition, eyes, eyebrow and mouth position
Textural characteristics be more advantageous to difference different people expression different faces texture information.
Referring to Fig. 4 and Fig. 5, the process that the topography containing face organ is extracted from the facial image can pass through
Following preferred embodiment is realized:
Step S211 determines the image-region in the facial image containing face organ according to preset ratio；
Since the human face ratio of different faces image is different, contain face organ in order to obtain in different faces image
Image-region, the preset ratio is preferably the ratio of the large range of image containing face organ of obtaining, such as: it is described
Preset ratio is the ratio greater than five traditional, three front yard, using the facial image upper left side as coordinate origin, eyes eyebrow
Image-region isThe image-region of mouth isThe w is described
The width of facial image, the h are the height of the facial image.
Step S212 extracts the topography containing eyes, eyebrow and/or mouth from described image region.
The topography that eyes eyebrow is extracted from the image-region of identified eyes eyebrow, from identified mouth
Mouth topography is extracted in image-region, such as: the size of eyes eyebrow topography isThe mouth
The size of bar topography isThe w is the width of the facial image, and the h is the height of the facial image.
Step S22, whole to the facial image and described topography carry out 2D-Gabor filtering processing respectively, generate
Multiple initial texture images.
The 2D-Gabor filtering processing can be one-dimensional 2D-Gabor filtering processing, be also possible to two-dimentional 2D-Gabor filter
Wave processing, the two-dimensional filtering processing are included in a scale of N (N is positive integer) a direction M (M is positive integer) and carry out 2D-Gabor
Filtering processing generates N*M initial texture image.In the present embodiment, two dimension 2D- preferably is carried out to the facial image
Gabor filtering processing.Such as: to the facial image entirety, the topography comprising eyes eyebrow and include the part of mouth
Image carries out the two-dimentional 2D-Gabor filtering processing of M, N number of direction scale respectively, generates the initial texture of N*M face entirety
The initial texture image of image, the initial texture image of N*M eyebrow eye and N*M mouth.
The present embodiment is grouped by the multiple initial texture image averagings for generating 2D-Gabor filtering processing, and from each group
Characteristic image is extracted in initial texture image, and passes through most significant and the face organ most beneficial for identification the part to extraction
Image carries out 2D-Gabor filtering processing, eliminates some unessential textural characteristics in initial texture image, so that extracting
Face textural characteristics it is more prominent, be conducive to identifying processing.
Above-mentioned first to second embodiment extraction face texture method executing subject all can be face texture mention
The face recognition device for taking equipment or being connect with face texture blending device signal.Further, the extraction face texture
Method can by be mounted on face texture blending equipment or face recognition device client detection program realize.
The present invention further provides a kind of devices for extracting face texture.
Referring to Fig. 6, Fig. 6 is a kind of functional module signal of the preferred embodiment for the device for extracting face texture of the present invention
Figure.
In the present embodiment, described device includes: read module 10, filter module 20, grouping module 30 and extraction module
40。
Read module 10, for reading facial image；
Filter module 20 generates multiple initial texture figures for carrying out 2D-Gabor filtering processing to the facial image
Picture；
The facial image of face textural characteristics to be extracted is read, the facial image can be carry out face recognition process
In facial image to be identified, be also possible to the sample image for comparison, or be also possible to other face textures to be extracted
Facial image.The initial texture image can be one or more than one initial texture image, it is preferred that scheme
As the form of matrix is indicated.The 2D-Gabor filtering processing can be one-dimensional 2D-Gabor filtering processing, be also possible to
Two-dimentional 2D-Gabor filtering processing, the two-dimensional filtering processing are included in N (N is positive integer) a direction M (M is positive integer) a ruler
Degree carries out 2D-Gabor filtering processing, generates N*M initial texture image.In the present embodiment, preferably to the face figure
As carrying out two dimension 2D-Gabor filtering processing.It should be understood that carrying out 2D-Gabor filtering processing to the face, and unlimited
In integrally carrying out 2D-Gabor filtering processing to the facial image, it is also possible to carry out 2D- to the topography of the face
Gabor filtering processing, or to the facial image is whole and the topography of the facial image carries out 2D-Gabor respectively
Filtering processing.Such as: it is whole to the facial image to carry out two dimension 2D-Gabor filtering processing in M, N number of direction scale, it generates
N*M initial texture image；Or the topography to the facial image, topography such as comprising eyes eyebrow and comprising
The topography of mouth, M, N number of direction, scale carries out two dimension 2D-Gabor filtering processing respectively, generates 2*N*M initial lines
Manage image；Or to the facial image entirety, the topography comprising eyes eyebrow and include the topography of mouth, respectively
The two-dimentional 2D-Gabor filtering processing for carrying out M, N number of direction scale, generates 3*N*M initial texture image.
Grouping module 30, for carrying out average packet to the multiple initial texture image；
Preferably, referring to Fig. 7, the grouping mould 30 includes sequencing unit 31 and division unit 32；The sequencing unit 31,
For the initial texture image of the same direction to be ranked up by scale；The division unit 32, for the same direction is adjacent
Two initial texture images of scale are divided into one group, successively carry out average packet.The initial texture image of the same direction is pressed
Scale is ranked up；Two initial texture images of the adjacent scale of the same direction are divided into one group, successively carry out average packet.
Two initial texture images of the adjacent scale of the same direction can be divided into one group, it can also be according to will be more than two initial
The mode that texture image is divided into one group carries out average packet.Such as: it is whole to the facial image a in N (N is positive integer)
A scale of direction M (M is positive integer) carries out two dimension 2D-Gabor filtering processing, N*M initial texture image is generated, by Xiang Tongfang
To initial texture image be ranked up by scale, two initial texture images of the adjacent scale of the same direction are divided into one
Group successively carries out average packet, and first initial texture image of all directions and second initial texture image are divided into one group,
The third initial texture image of all directions and the 4th initial texture image are divided into one group, and so on obtain N* (M/2)
A initial texture image group；Alternatively, topography and packet to the topography of the facial image, such as comprising eyes eyebrow
Topography containing mouth, M, N number of direction, scale carries out two dimension 2D-Gabor filtering processing respectively, generates 2*N*M initially
Two initial texture images of the adjacent scale of the same direction of the initial texture image of eyes eyebrow are divided into one by texture image
Group successively carries out average packet, a eyes eyebrow initial texture image group of N* (M/2) is obtained, by the initial texture image of mouth
Two initial texture images of the adjacent scale of the same direction be divided into one group, successively carry out average packet, it is a to obtain N* (M/2)
Mouth initial texture image group；Alternatively, to the facial image entirety, the topography comprising eyes eyebrow and including mouth
Topography carries out the two-dimentional 2D-Gabor filtering processing of M, N number of direction scale respectively, generates the initial of N*M face entirety
The initial texture image of texture image, the initial texture image of N*M eyes eyebrow and N*M mouth, will face it is whole just
Two initial texture images of the adjacent scale of the same direction of beginning texture image are divided into one group, successively carry out average packet, obtain
To the initial texture image group of a face entirety of N* (M/2), by the adjacent ruler of the same direction of the initial texture image of eyes eyebrow
Two initial texture images of degree are divided into one group, successively carry out average packet, obtain a eyes eyebrow initial texture of N* (M/2)
Two initial texture images of the adjacent scale of the same direction of the initial texture image of mouth are divided into one group by image group, according to
Secondary carry out average packet obtains a mouth initial texture image group of N* (M/2).
Extraction module 40 for extracting characteristic image from every group of initial texture image respectively, and exports the characteristic pattern
Picture.
Extract most the characteristic image that can represent face textural characteristics from every group of initial texture image respectively, and by institute
State characteristic image output.Such as: to the facial image entirety, the topography comprising eyes eyebrow and include the part of mouth
Image carries out the two-dimentional 2D-Gabor filtering processing of N (N is positive integer) a direction M (M is positive integer) a scale respectively, generates
The initial texture of the initial texture image of N*M face entirety, the initial texture image of N*M eyes eyebrow and N*M mouth
Image is divided into according to two initial texture images by the adjacent scale of the same direction one group of side to the initial texture image
Formula carries out average packet, at the beginning of respectively obtaining the initial texture image group of a face entirety of N* (M/2), (N*M)/2 eyes eyebrow
Beginning texture image group and a mouth initial texture image group of N* (M/2), extract most from every group of initial texture image respectively
The characteristic image that face textural characteristics can be represented obtains the characteristic image of a face entirety of N* (M/2), a eyes eyebrow of N* (M/2)
The characteristic image of hair and the characteristic image of N* (M/2) a mouth.It is the preferred embodiment of the extraction module 40 referring to Fig. 8, Fig. 8
The functional block diagram, the extraction module 40 include decomposition unit 41, determination unit 42 and assembled unit 43；
Decomposition unit 41, for each initial texture image averaging in each group to be decomposed into image block；
Initial texture image in each group can be equally divided into i (i is positive integer) part, width direction in the height direction
On be equally divided into j (j is positive integer) part, each initial texture picture breakdown obtains i*j image block, and the short transverse is institute
The vertical direction of initial texture image is stated, the width direction is the horizontal direction of the initial texture image, it is preferred that will be every
The i*j image block that a initial texture picture breakdown obtains is indicated with the image array that i row j is arranged.
Determination unit 42, for determining the maximum image block of same position pixel value in every group, as characteristic image block；
Determine the pixel value of each image block, it is preferred that determine that the process of the pixel value of each image block can be, one by one
The pixel value for inquiring each pixel of described image block, takes the pixel value of the maximum pixel of pixel value as described image block
Pixel value.By in every group, the pixel value of the image block of same position is compared, and takes the maximum image block of pixel value as feature
Image block.Such as: have the first initial texture image and the second initial texture image in an initial texture image group, by this two
A initial texture image averaging is decomposed into i*j (i is positive integer, and j is positive integer) a image block, that is, by initial texture image
Height is equally divided into i equal portions, and the width of initial texture image is equally divided into j equal portions.The pixel value for determining each image block, is looked into one by one
The pixel value for asking each pixel of described image block, takes pixel of the pixel value of the maximum point of pixel value as described image block
Value.I-th row jth of the pixel value of the i-th row jth column image block of the first initial texture image and the second initial texture image is arranged
The pixel value of image block compares, and takes the maximum image block of pixel value as characteristic image block.Every group obtains i*j characteristic pattern
As block.
Assembled unit 43 for every group of characteristic image block to be combined into characteristic image, and exports the characteristic image.
By the characteristic image block in every group by preset order permutation and combination at characteristic image, and export the characteristic image.
Such as: the characteristic image block extracted is arranged from i-th (i is positive integer) row jth (j is positive integer) of initial texture image, then
Come characteristic image the i-th row jth column, and so on carry out characteristic image combination, using the characteristic image as representative
The image of face textural characteristics exports.
The present embodiment is grouped by the multiple initial texture image averagings for generating 2D-Gabor filtering processing, and from each group
Characteristic image is extracted in initial texture image, some unessential textural characteristics in initial texture image are eliminated, so that mentioning
The face textural characteristics taken are more prominent, are conducive to identifying processing.
It is the functional block diagram of the second embodiment for the device that the present invention extracts face texture referring to Fig. 9, Fig. 9.Base
In the first embodiment of the device of said extracted face texture, the filter module 20 includes acquiring unit 21 and filter unit
22；
The acquiring unit 21, for extracting the topography containing face organ from the facial image；
Facial image is read, the topography containing face organ, the face organ are extracted from the facial image
Including eyes, eyebrow, nose or mouth etc..It can be extracted from the facial image containing eyes, eyebrow, nose or mouth
The topography of equal face organs' one of which is handled；It can also be extracted from the facial image containing eyes, eyebrow
More than one the topography of the face organs such as hair, mouth or nose is handled.Preferably, it is mentioned from the facial image
Take out the topography containing eyes eyebrow, mouth, during carrying out facial expression recognition, eyes, eyebrow and mouth position
Textural characteristics be more advantageous to difference different people expression different faces texture information.
The filter unit 22, for whole to the facial image and the topography to carry out 2D-Gabor filter respectively
Wave processing, generates multiple initial texture images.
Since the human face ratio of different faces image is different, contain face organ in order to obtain in different faces image
Image-region, the preset ratio is preferably the ratio of the large range of image containing face organ of obtaining, such as: it is described
Preset ratio is the ratio greater than five traditional, three front yard, using the facial image upper left side as coordinate origin, eyes eyebrow
Image-region isThe image-region of mouth isThe w is described
The width of facial image, the h are the height of the facial image.
The acquiring unit 21 is also used to determine the image in the facial image containing face organ according to preset ratio
Region；
The topography that eyes eyebrow is extracted from the image-region of identified eyes eyebrow, from identified mouth
Mouth topography is extracted in image-region, such as: the size of eyes eyebrow topography isThe mouth
The size of bar topography isThe w is the width of the facial image, and the h is the height of the facial image.
The acquiring unit 21 is also used to extract the part containing eyes, eyebrow and/or mouth from described image region
Image.
The 2D-Gabor filtering processing can be one-dimensional 2D-Gabor filtering processing, be also possible to two-dimentional 2D-Gabor filter
Wave processing, the two-dimensional filtering processing are included in a scale of N (N is positive integer) a direction M (M is positive integer) and carry out 2D-Gabor
Filtering processing generates N*M initial texture image.In the present embodiment, two dimension 2D- preferably is carried out to the facial image
Gabor filtering processing.Such as: to the facial image entirety, the topography comprising eyes eyebrow and include the part of mouth
Image carries out the two-dimentional 2D-Gabor filtering processing of M, N number of direction scale respectively, generates the initial line of N*M face entirety
Manage the initial texture image of image, the initial texture image of N*M eyebrow eye and N*M mouth.
The present embodiment is grouped by the multiple initial texture image averagings for generating 2D-Gabor filtering processing, and from each group
Characteristic image is extracted in initial texture image, and passes through most significant and the face organ most beneficial for identification the part to extraction
Image carries out 2D-Gabor filtering processing, eliminates some unessential textural characteristics in initial texture image, so that extracting
Face textural characteristics it is more prominent, be conducive to identifying processing.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.Pass through above embodiment party
The description of formula, it is required general that those skilled in the art can be understood that above-described embodiment method can add by software
The mode of hardware platform is realized, naturally it is also possible to which by hardware, but in many cases, the former is more preferably embodiment.It is based on
Such understanding, substantially the part that contributes to existing technology can be with software product in other words for technical solution of the present invention
Form embody, which is stored in a storage medium (such as ROM/RAM, magnetic disk, CD), including
Some instructions are used so that a terminal device (can be mobile phone, computer, server or the network equipment etc.) executes this hair
Method described in bright each embodiment.
The above is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (8)
1. a kind of method for extracting face texture, which is characterized in that the method for extracting face texture includes the following steps:
Read facial image；
2D-Gabor filtering processing is carried out to the facial image, generates multiple initial texture images；
Average packet is carried out to the multiple initial texture image；
Each initial texture image averaging in each group is decomposed into image block；
The maximum image block of same position pixel value in every group is determined, as characteristic image block；
Every group of characteristic image block is combined into characteristic image, and exports the characteristic image.
2. extracting the method for face texture as described in claim 1, which is characterized in that described to be carried out to the facial image
2D-Gabor filtering processing, the step of generating multiple initial texture images include:
The topography containing face organ is extracted from the facial image；
Whole to the facial image and described topography carries out 2D-Gabor filtering processing respectively, generates multiple initial textures
Image.
3. extracting the method for face texture as claimed in claim 2, which is characterized in that described to be extracted from the facial image
The step of topography containing face organ includes:
The image-region in the facial image containing face organ is determined according to preset ratio；
The topography containing eyes, eyebrow and/or mouth is extracted from described image region.
4. extracting the method for face texture as described in claim 1, which is characterized in that described to the multiple initial texture figure
As the step of carrying out average packet includes:
The initial texture image of the same direction is ranked up by scale；
Two initial texture images of the adjacent scale of the same direction are divided into one group, successively carry out average packet.
5. it is a kind of extract face texture device, which is characterized in that it is described extract face texture device include:
Read module, for reading facial image；
Filter module generates multiple initial texture images for carrying out 2D-Gabor filtering processing to the facial image；
Grouping module, for carrying out average packet to the multiple initial texture image；
Extraction module, including decomposition unit, determination unit and assembled unit；
The decomposition unit, for each initial texture image averaging in each group to be decomposed into image block；
The determination unit, for determining the maximum image block of same position pixel value in every group, as characteristic image block；
The assembled unit for every group of characteristic image block to be combined into characteristic image, and exports the characteristic image.
6. extracting the device of face texture as claimed in claim 5, which is characterized in that the filter module includes acquiring unit
And filter unit；
The acquiring unit, for extracting the topography containing face organ from the facial image；
The filter unit, for whole to the facial image and the topography to carry out at 2D-Gabor filtering respectively
Reason, generates multiple initial texture images.
7. as claimed in claim 6 extract face texture device, which is characterized in that the acquiring unit, be also used to according to
Preset ratio determines the image-region in the facial image containing face organ；
The acquiring unit is also used to extract the topography containing eyes, eyebrow and/or mouth from described image region.
8. extracting the device of face texture as claimed in claim 5, which is characterized in that the grouping module includes sequencing unit
And division unit；
The sequencing unit, for the initial texture image of the same direction to be ranked up by scale；
The division unit is successively carried out for two initial texture images of the adjacent scale of the same direction to be divided into one group
Average packet.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510191874.4A CN105678208B (en) | 2015-04-21 | 2015-04-21 | Method and device for extracting face texture |
| PCT/CN2015/091503 WO2016169219A1 (en) | 2015-04-21 | 2015-10-09 | Method and device for extracting human facial textures |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510191874.4A CN105678208B (en) | 2015-04-21 | 2015-04-21 | Method and device for extracting face texture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105678208A CN105678208A (en) | 2016-06-15 |
| CN105678208B true CN105678208B (en) | 2019-03-08 |
Family
ID=56946809
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510191874.4A Active CN105678208B (en) | 2015-04-21 | 2015-04-21 | Method and device for extracting face texture |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN105678208B (en) |
| WO (1) | WO2016169219A1 (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107392183B (en) * | 2017-08-22 | 2022-01-04 | 深圳Tcl新技术有限公司 | Face classification recognition method and device and readable storage medium |
| CN110046587B (en) * | 2019-04-22 | 2022-11-25 | 安徽理工大学 | Facial expression feature extraction method based on Gabor differential weight |
| CN110569873A (en) * | 2019-08-02 | 2019-12-13 | 平安科技（深圳）有限公司 | Image recognition model training method and device and computer equipment |
| CN112733570B (en) * | 2019-10-14 | 2024-04-30 | 北京眼神智能科技有限公司 | Glasses detection method and device, electronic equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1908960A (en) * | 2005-08-02 | 2007-02-07 | 中国科学院计算技术研究所 | Feature classification based multiple classifiers combined people face recognition method |
| CN102254304A (en) * | 2011-06-17 | 2011-11-23 | 电子科技大学 | Method for detecting contour of target object |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101414348A (en) * | 2007-10-19 | 2009-04-22 | 三星电子株式会社 | Method and system for identifying human face in multiple angles |
| CN101763507B (en) * | 2010-01-20 | 2013-03-06 | 北京智慧眼科技发展有限公司 | Face recognition method and face recognition system |
| CN102750523B (en) * | 2012-06-19 | 2016-04-06 | Tcl集团股份有限公司 | A kind of method of recognition of face and device |
| CN103902977B (en) * | 2014-03-31 | 2017-04-05 | 华为技术有限公司 | Face identification method and device based on Gabor binary patterns |
-
2015
- 2015-04-21 CN CN201510191874.4A patent/CN105678208B/en active Active
- 2015-10-09 WO PCT/CN2015/091503 patent/WO2016169219A1/en active Application Filing
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1908960A (en) * | 2005-08-02 | 2007-02-07 | 中国科学院计算技术研究所 | Feature classification based multiple classifiers combined people face recognition method |
| CN102254304A (en) * | 2011-06-17 | 2011-11-23 | 电子科技大学 | Method for detecting contour of target object |
Non-Patent Citations (2)
| Title |
|---|
| 基于环形对称Gabor变换和分块PCA的人脸识别;王甜甜 等;《电视技术》;20130802;173-176 * |
| 王甜甜 等.基于环形对称Gabor变换和分块PCA的人脸识别.《电视技术》.2013,173-176. * |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2016169219A1 (en) | 2016-10-27 |
| CN105678208A (en) | 2016-06-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Fang et al. | Bottom-up saliency detection model based on human visual sensitivity and amplitude spectrum | |
| CN104933414B (en) | A kind of living body faces detection method based on WLD-TOP | |
| CN105744256B (en) | Based on the significant objective evaluation method for quality of stereo images of collection of illustrative plates vision | |
| US11900557B2 (en) | Three-dimensional face model generation method and apparatus, device, and medium | |
| CN103456013B (en) | A kind of method representing similarity between super-pixel and tolerance super-pixel | |
| CN105678208B (en) | Method and device for extracting face texture | |
| CN109214298B (en) | Asian female color value scoring model method based on deep convolutional network | |
| CN104392233B (en) | A kind of image saliency map extracting method based on region | |
| CN104658030B (en) | The method and apparatus of secondary image mixing | |
| DE102018003475A1 (en) | Form-based graphic search | |
| Casanova et al. | Texture analysis using fractal descriptors estimated by the mutual interference of color channels | |
| Liu et al. | Blind stereoscopic image quality assessment based on hierarchical learning | |
| CN109903392A (en) | Augmented reality method and apparatus | |
| CN104137144B (en) | Image procossing digital filter and character string inclination illusion generating means | |
| CN111275784A (en) | Method and device for generating image | |
| CN104834909B (en) | A kind of new image representation method based on Gabor comprehensive characteristics | |
| Muthevi et al. | Leaf classification using completed local binary pattern of textures | |
| CN108388889A (en) | Method and apparatus for analyzing facial image | |
| Agarwal et al. | MD-CSDNetwork: Multi-domain cross stitched network for deepfake detection | |
| CN108182457A (en) | For generating the method and apparatus of information | |
| CN105528616B (en) | Face identification method and device | |
| CN113808277A (en) | Image processing method and related device | |
| CN109492601A (en) | Face comparison method and device, computer-readable medium and electronic equipment | |
| CN113673465B (en) | Image detection method, device, equipment and readable storage medium | |
| CN114202723A (en) | Intelligent editing application method, device, equipment and medium through picture 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |