CN105608448B

CN105608448B - A kind of LBP feature extracting method and device based on face's key point

Info

Publication number: CN105608448B
Application number: CN201610093896.1A
Authority: CN
Inventors: 冯谨强; 高伟杰
Original assignee: Hisense Group Co Ltd
Current assignee: Hisense Group Co Ltd
Priority date: 2016-02-22
Filing date: 2016-02-22
Publication date: 2019-08-09
Anticipated expiration: 2036-02-22
Also published as: CN105608448A

Abstract

The invention discloses a kind of LBP feature extracting methods and device based on face's key point, the present invention relates to Image Processing and Pattern Recognition technical fields, it aims to solve the problem that under same people, different faces posture, the larger problem of the LBP characteristic difference of the same key point of the face of extraction, and then improve face recognition accuracy rate.This method determines the human face posture of the face image according to the key point of face image, and then according to the human face posture of the face image correct different face's key points corresponding to Ellipse Neighborhood radius, establish the corresponding relationship between Ellipse Neighborhood radius and the human face posture of the face image, so as to avoid under same person's different faces posture, face image local scale ratio is different, the larger problem of the LBP characteristic difference for causing same key point to be extracted, improve the validity of the LBP feature extraction under different faces posture based on face's key point, and then improve face recognition accuracy rate.

Description

A kind of LBP feature extracting method and device based on face's key point

Technical field

The present invention relates to Image Processing and Pattern Recognition technical field more particularly to a kind of LBP based on face's key point Feature extracting method and device.

Background technique

Face recognition technology is to carry out identity identification by the displacement shape and positional relationship of analysis face's organ, is A kind of important biological identification technology, is widely used in the fields such as security protection, gate inhibition and monitoring.The main calculation of face recognition technology Method includes the face identification method of the template matching based on geometrical characteristic, the face identification method based on geometrical characteristic, is based on sample The face identification method of this study and face identification method based on textural characteristics.Wherein, the people based on face textural characteristics Face recognition method relies primarily on LBP (Local Binary Pattern) i.e. local binary patterns and carries out face feature extraction.

Facial feature extraction method based on LBP mainly includes two kinds, one is for entire face image carry out piecemeal, Multiple dimensioned LBP feature extraction is carried out to each block of image, combines all pieces of LBP feature to obtain entire face after normalization LBP feature；Another kind is extracted around it centered on face's key point (exemplary, eyeball, nose, the corners of the mouth, eyebrow etc.) The LBP feature of certain area, by the LBP feature of all key points after normalization, combination obtains the LBP feature of entire face.So And the above two facial feature extraction method based on LBP is with certain point to the face image of same person's difference posture Centered on point, acquire the pixel value of several sampled points in the round neighborhood of some radii fixus, and then to several samplings The pixel value of point is compared with central point pixel value, obtains the LBP characteristic value of the central point.Therefore, existing based on LBP's Facial feature extraction method does not account for the difference of face image scaling caused by attitudes vibration, leads to the same person Face image in different positions, it is several in the circle shaped neighborhood region of extracted same radius for the central point of its same position The pixel value of a sampled point is different, and the LBP feature of the central point is caused to differ larger, and then causes recognition of face inaccurate.

It is exemplary, as shown in Figure 1 and Figure 2, when being respectively the positive face of the same person and left side of the face shown in Fig. 1 and Fig. 2 Face image, the circle that the LBP feature extraction for central point A same at its corners of the mouth, first radius shown in extraction Fig. 1, Fig. 2 are R The pixel value of N number of sampled point in shape neighborhood, referring to fig. 1 and fig. 2, when due to left side of the face, radius is N in the circle shaped neighborhood region of R Fractional-sample point in a sampled point has exceeded face range, leads to the identical circle shaped neighborhood region of same central point A shown in Fig. 1, Fig. 2 The pixel value difference of interior N number of sampled point is larger, so cause Fig. 1, Fig. 2 propose place same central point A LBP characteristic difference It is larger, and then cause face image shown in Fig. 1, Fig. 2 that can not be identified as same people, cause recognition of face inaccurate.

Summary of the invention

The embodiment of the present invention provides a kind of facial feature extraction method and device based on LBP, it is intended to solve same people not With under human face posture, the larger problem of the LBP characteristic difference of the same key point of the face of extraction, and then it is accurate to improve recognition of face Rate.

Specific technical solution provided by the invention is as follows:

A kind of LBP feature extracting method based on face's key point, comprising:

Based on the key point of face image to be processed, the human face posture of the face image is estimated；

According to the human face posture of the face image, the corresponding Ellipse Neighborhood radius of the key point is corrected；

LBP feature extraction is carried out to the key point using the Ellipse Neighborhood radius to the face image.

Further, the key point based on face image to be processed estimates the human face posture of the face image, packet It includes:

Obtain the first vertical distance and the second vertical distance on the face image vertical direction, wherein described first Vertical distance and second vertical distance are distributed from up to down along the vertical direction；

Ratio between first vertical distance and second vertical distance is greater than first threshold, then face's figure The human face posture of picture is to bow；Or the ratio between first vertical distance and second vertical distance is less than the first threshold Value, then the human face posture of the face image is to face upward head.

Further, the key point based on face image to be processed, estimates the human face posture of the face image, also Include:

Obtain the first level distance and the second horizontal distance in the face image horizontal direction, wherein described first Horizontal distance and second horizontal distance are distributed from left to right in the horizontal direction；

Ratio between the first level distance and second horizontal distance is greater than second threshold, then face's figure The human face posture of picture is right side head；Or the ratio between the first level distance and second horizontal distance is less than the second threshold Value, then the human face posture of the face image is left side head.

Preferably, the human face posture according to the face image corrects the corresponding Ellipse Neighborhood of the key point half Diameter, comprising:

The horizontal axis of the face if the human face posture is positive, the Ellipse Neighborhood is equal to the first preset value, the Ellipse Neighborhood The longitudinal axis is equal to the first preset value；If or the human face posture is faces upward head or bow, it is pre- that the horizontal axis of the Ellipse Neighborhood is equal to first If value, the longitudinal axis of the Ellipse Neighborhood is equal to the product of the first preset value and the first default weight；If or the human face posture is Left side head or right side head, the horizontal axis of the Ellipse Neighborhood are equal to the product of the first preset value and the second default weight, the ellipse The longitudinal axis of neighborhood is equal to the first preset value.

Preferably, the face image includes first area, second area, third region and the fourth region, wherein described First area, the second area, the third region and the fourth region are distributed along clockwise, and first area position In on the fourth region vertical direction.

Further, the human face posture according to the face image corrects the corresponding Ellipse Neighborhood of the key point Radius, comprising:

The face if human face posture is positive, the first area, the second area, the third region and the described 4th The horizontal axis of the Ellipse Neighborhood in region and the longitudinal axis of the Ellipse Neighborhood are equal to the first preset value；

If the human face posture is to face upward head, the first area, the second area, the third region and the described 4th The horizontal axis in region is equal to first preset value, and the longitudinal axis of the first area and the second area is less than first The longitudinal axis of preset value, the third region and the fourth region is greater than first preset value；

If the human face posture is to bow, the first area, the second area, the third region and the described 4th The horizontal axis in region is equal to first preset value, and the longitudinal axis of the first area and the second area is greater than first The longitudinal axis of preset value, the third region and the fourth region is less than first preset value；

If the human face posture is left side head, it is pre- that the horizontal axis in the first area and the third region is greater than first If value, the longitudinal axis of the second area and the fourth region is less than first preset value, the first area, described The longitudinal axis of second area, the third region and the fourth region is equal to first preset value；

If the human face posture is right side head, the horizontal axis in the first area and the third region is pre- less than first If value, the longitudinal axis of the second area and the fourth region is greater than first preset value, the first area, described The longitudinal axis of second area, the third region and the fourth region is equal to first preset value；

If the human face posture is to face upward head and left side head, the horizontal axis in the first area and the third region is greater than The horizontal axis of first preset value, the second area and the fourth region is less than first preset value, and described the The longitudinal axis of one region and the second area less than the first preset value, the third region and the fourth region it is described The longitudinal axis is greater than the first preset value；

If the human face posture is to face upward head and right side head, the horizontal axis in the first area and the third region is less than The horizontal axis of first preset value, the second area and the fourth region is greater than first preset value, and described the The longitudinal axis of one region and the second area is greater than the first preset value, the third region and the fourth region it is described The longitudinal axis is less than the first preset value；

If the human face posture is to bow and left side head, the horizontal axis in the first area and the third region is greater than The horizontal axis of first preset value, the second area and the fourth region is less than first preset value, and described the The longitudinal axis of one region and the second area is greater than the first preset value, the third region and the fourth region it is described The longitudinal axis is less than the first preset value；

If the human face posture is to face upward head and left side head, the horizontal axis in the first area and the third region is less than The horizontal axis of first preset value, the second area and the fourth region is greater than first preset value, and described the The longitudinal axis of one region and the second area less than the first preset value, the third region and the fourth region it is described The longitudinal axis is greater than the first preset value.

On the other hand, the present invention also provides a kind of LBP feature extracting methods based on face's key point, comprising:

The key point corresponding Ellipse Neighborhood in preset mapping table is determined according to the human face posture of the face image Radius；

Preferably, the preset mapping table is used to characterize the key under different faces posture, in different face areas The corresponding Ellipse Neighborhood radius of point, in which:

On the one hand, the present invention also provides a kind of LBP feature deriving means based on face's key point, comprising:

Critical point detection module, for detecting the key point of face image to be processed；

Image processing module determines that the face schemes according to the distance for obtaining the distance between described key point The human face posture of picture, and the radius that the key point corresponds to Ellipse Neighborhood is corrected according to the human face posture；

Characteristic extracting module, for being carried out using the radius of the Ellipse Neighborhood to the key point to the face image LBP feature extraction.

Further, described image processing module is specifically used for:

Obtain the first vertical distance and the second vertical distance on the face image vertical direction, wherein described first Vertical distance and second vertical distance are distributed from up to down along the vertical direction, and first vertical distance is perpendicular with described second Ratio between straight distance is greater than first threshold, then the human face posture of the face image be bow or described first vertically away from It is less than first threshold from the ratio between second vertical distance, then the human face posture of the face image is to face upward head；

Obtain the first level distance and the second horizontal distance in the face image horizontal direction, wherein described first Horizontal distance and second horizontal distance are distributed from left to right in the horizontal direction, the first level distance and second water Ratio between flat distance is greater than second threshold, then the human face posture of the face image is right side head or the first level Ratio between distance and second horizontal distance is less than second threshold, then the human face posture of the face image is left side Head.

Further, described image processing module is specifically used for:

The face if human face posture is positive determines the first area, the second area, the third region and described The horizontal axis of the Ellipse Neighborhood of the fourth region and the longitudinal axis of the Ellipse Neighborhood are equal to the first preset value；Or

If the human face posture is to face upward head, the first area, the second area, the third region and described are determined The horizontal axis of the fourth region is equal to first preset value, and the longitudinal axis of the first area and the second area is less than The longitudinal axis of first preset value, the third region and the fourth region is greater than first preset value；Or

If the human face posture is to bow, the first area, the second area, the third region and described are determined The horizontal axis of the fourth region is equal to first preset value, and the longitudinal axis of the first area and the second area is greater than The longitudinal axis of first preset value, the third region and the fourth region is less than first preset value；Or

If the human face posture is left side head, determine that the horizontal axis in the first area and the third region is greater than the The longitudinal axis of one preset value, the second area and the fourth region be less than first preset value, the first area, The longitudinal axis of the second area, the third region and the fourth region is equal to first preset value；Or

If the human face posture is right side head, determine the horizontal axis in the first area and the third region less than the The longitudinal axis of one preset value, the second area and the fourth region be greater than first preset value, the first area, The longitudinal axis of the second area, the third region and the fourth region is equal to first preset value；Or

If the human face posture is to face upward head and left side head, the horizontal axis of the first area and the third region is determined Greater than first preset value, the horizontal axis of the second area and the fourth region is less than first preset value, institute The longitudinal axis of first area and the second area is stated less than the first preset value, the third region and the fourth region The longitudinal axis is greater than the first preset value；Or

If the human face posture is to face upward head and right side head, the horizontal axis of the first area and the third region is determined Less than first preset value, the horizontal axis of the second area and the fourth region is greater than first preset value, institute The longitudinal axis for stating first area and the second area is greater than the first preset value, the third region and the fourth region The longitudinal axis is less than the first preset value；Or

If the human face posture is to bow and left side head, the horizontal axis of the first area and the third region is determined Greater than first preset value, the horizontal axis of the second area and the fourth region is less than first preset value, institute The longitudinal axis for stating first area and the second area is greater than the first preset value, the third region and the fourth region The longitudinal axis is less than the first preset value；Or

If the human face posture is to face upward head and left side head, the horizontal axis of the first area and the third region is determined Less than first preset value, the horizontal axis of the second area and the fourth region is greater than first preset value, institute The longitudinal axis of first area and the second area is stated less than the first preset value, the third region and the fourth region The longitudinal axis is greater than the first preset value.

Beneficial effects of the present invention are as follows:

LBP feature extracting method provided in an embodiment of the present invention based on face's key point, first according to face image Key point determines the human face posture of the face image, and then corrects different face's key points according to the human face posture of the face image Corresponding Ellipse Neighborhood radius, establishes the corresponding relationship between Ellipse Neighborhood radius and the human face posture of the face image, So as to avoid under same person's different faces posture, face image local scale ratio is different, and same key point is caused to be extracted The larger problem of LBP characteristic difference, improve the effective of the LBP feature extraction under different faces posture based on face's key point Property, and then improve face recognition accuracy rate.

Detailed description of the invention

The LBP feature extraction neighborhood schematic diagram that Fig. 1 is face key point A under positive face posture in the prior art；

The LBP feature extraction neighborhood schematic diagram that Fig. 2 is face key point A under side face posture in the prior art；

Fig. 3 is a kind of LBP feature extraction flow diagram based on face's key point of the embodiment of the present invention；

Fig. 4 is a kind of face's key point distribution schematic diagram of the embodiment of the present invention；

Fig. 5 is the face image schematic diagram to be processed before a kind of image rectification of the embodiment of the present invention；

Fig. 6 is the face image schematic diagram to be processed after a kind of image rectification of the embodiment of the present invention

Fig. 7 is a kind of face's face position distribution schematic diagram of the embodiment of the present invention；

Fig. 8 is a kind of human face modeling face's key point distribution schematic diagram of the embodiment of the present invention；

Fig. 9 is that a kind of Ellipse Neighborhood of the embodiment of the present invention corrects schematic diagram；

Figure 10 is that a kind of face area of the embodiment of the present invention divides schematic diagram；

Figure 11 is another LBP feature extraction flow diagram based on face's key point of the embodiment of the present invention；

Figure 12 is a kind of LBP feature deriving means schematic diagram based on face's key point of the embodiment of the present invention.

Specific embodiment

To make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with attached drawing to the present invention make into It is described in detail to one step, it is clear that described embodiments are only a part of the embodiments of the present invention, rather than whole implementation Example.Based on the embodiments of the present invention, obtained by those of ordinary skill in the art without making creative efforts All other embodiment, shall fall within the protection scope of the present invention.

In the description of the present invention, it is to be understood that, term " center ", "upper", "lower", "front", "rear", " left side ", The orientation or positional relationship of the instructions such as " right side ", "vertical", "horizontal", "top", "bottom", "inner", "outside" is based on the figure Orientation or positional relationship is merely for convenience of description of the present invention and simplification of the description, rather than the device of indication or suggestion meaning or Element must have a particular orientation, be constructed and operated in a specific orientation, therefore be not considered as limiting the invention.

Term " first ", " second ", " third ", " the 4th " " the 5th ", " the 6th ", " the 7th ", " the 8th " and " the 9th " is only For descriptive purposes, it is not understood to indicate or imply relative importance or implicitly indicates the number of indicated technical characteristic Amount." first ", " second ", " third ", " the 4th " " the 5th ", " the 6th ", " the 7th ", " the 8th " and " the 9th " are defined as a result, Feature can explicitly or implicitly include one or more of the features.In the description of the present invention, unless otherwise saying Bright, the meaning of " plurality " is two or more.

The embodiment of the present invention can be applied to various intelligent terminals, as smart television, mobile phone, intelligent video camera head, monitoring are set It is standby etc.；The embodiment of the present invention is particularly suitable for each Terminal Type with face recognition technology, for example, by using face recognition technology Security device, monitoring device, access control equipment, smart television and smart phone etc..

Embodiment one

Fig. 3 shows a kind of LBP feature extraction flow diagram based on face's key point provided in an embodiment of the present invention, As shown in figure 3, the LBP characteristic extraction procedure based on face's key point includes:

Step 100: the key point based on face image to be processed estimates the human face posture of the face image.

Specifically, face image to be processed is pre-processed first during executing step 100, obtain to The gray level image for handling image carries out face's critical point detection to the gray level image of the image to be processed, obtains the gray level image In LBP feature to be extracted face's key point position.It should be noted that determining face's key point includes in step 100 Face face exterior feature key point, eyebrow key point, eyes key point, nose key point and mouth key point, respectively represent face to be processed The position of the position of facial contour, the position of eyebrow, the position of eyes, the position of nose and mouth in portion's image.

Exemplary, Fig. 4 shows a kind of face's key point distribution schematic diagram of the embodiment of the present invention, as shown in figure 4, determining Face's key point share 83, wherein the facial contour key point for representing face outline position and facial contour size has 19 A, the eyebrow key point for representing eyebrow position and eyebrow size has 16, and the eyes for representing eye position and eyes size are crucial Point has 18, and the nose key point for representing nose shape and nose size has 12, represents the mouth of mouth position and mouth size Bar key point has 18.Certainly, it is merely illustrative of herein, does not represent the face that the step 100 of the embodiment of the present invention determines Key point distributing position and quantity are confined to this.

Further, it should be noted that face's key point position of face image to be processed has been determined in step 100 Later, it needs face image to be processed integrally carrying out image rectification, to ensure in each face image to be processed, two eyeballs Between line and horizontal line between angle it is identical.It is exemplary, it can guarantee institute by rotating entire face image to be processed In pending face image, the ordinate for representing the key point of two eyeball position of face is identical, i.e., at guarantee human eye eyeball In in same horizontal line.

It is exemplary, refering to what is shown in Fig. 5, representing two eyeball of face before face image to be processed does not carry out image rectification There is certain angle between line and horizontal line between the key point of position；Face image described in reference diagram 6, to be processed After carrying out image rectification, the line represented between the key point of two eyeball position of face is in same horizontal line, that is, is represented The ordinate of the key point of two eyeball position of face is identical；With reference to shown in Fig. 5 and Fig. 6, figure is carried out to face image to be processed As correction, only in plane coordinate system, face image to be processed is integrally rotated, does not change face's figure to be processed Human face posture and face size as in.

Further, with reference to a large amount of experience, human face posture generally includes positive face, left side of the face, right side face, bows and face upward head Deng five kinds of different postures, wherein left side of the face and right side face are relative to positive face, and the face feature of face image to be processed is in level It changes greatly on direction, does not change in the vertical direction；Head is bowed and faced upward relative to positive face, the face of face image to be processed Portion's feature does not change in the horizontal direction, changes greatly in the vertical direction；Based on above-mentioned analysis, the execution of step 100 Process includes:

Specifically, obtaining the first key point to the first vertical distance and second key point between the second key point To the second vertical distance between third key point, wherein first key point, second key point and the third Key point is distributed from up to down along the vertical direction；

Third key point is obtained to the first level distance and the 5th key point to the 6th key between the 4th key point The second horizontal distance between point, wherein the third key point, the 4th key point, the 5th key point and institute The 6th key point is stated from left to right to be distributed in the horizontal direction；

It should be noted that judging whether human face posture is to face upward head or bow and judge people in the implementation procedure of step 100 Whether face posture is that sequencing is not present between left side head and right side head, and the two can be interchanged sequence, can also hold simultaneously Row.I.e. the implementation procedure of step 100 can also be as follows:

Further, it should be noted that the second vertical distance can be the second key point to erecting between third key point Straight distance, or third key point to the vertical distance between the 9th key point, wherein the first key point, the second key Point, third key point and the 9th key point are distributed from up to down along the vertical direction.

Further, it should be noted that the 4th key point and the 5th key point can be the same pass in facial image Key point is also possible to the different key points being distributed in the horizontal direction on facial image.

Further, it should be noted that first threshold and second threshold respectively represent under positive face-like state in face image One vertical distance and the ratio of the second vertical distance and the ratio of first level distance and the second horizontal distance, first threshold and the Two threshold values can be respectively a specific value, can also be respectively a numberical range.

Below in conjunction with specific embodiments, step 100 implementation procedure of the embodiment of the present invention is described in detail, certainly, It is merely illustrative of herein, the human face modeling for not representing the embodiment of the present invention is confined to this.

Fig. 7 illustratively shows positive face posture human face face position distribution schematic diagram, refering to what is shown in Fig. 7, eyeball arrives Vertical distance and eyeball between volume baseline to the vertical distance between the highest point of face head accounted for respectively face vertically away from From half, i.e., the line between two eyeballs of face is center line of the face in vertical direction, wherein nose baseline to volume baseline Between vertical distance and nose baseline accounted for the one third of face vertical distance respectively to the vertical distance between open wiring, i.e., Under positive face posture, the ratio of open wiring to the vertical distance between nose baseline and nose baseline to the vertical distance between volume baseline is 1: 1.Further analysis finds, when human face posture is become facing upward from positive face, open wiring compares nose to the face part between nose baseline The distance of baseline to the face partial distance taking lens between volume baseline is remote, corresponding open wiring between nose baseline it is vertical away from From becoming larger under relatively positive face-like state, becomes smaller under the relatively positive face-like state of nose baseline to the vertical distance between volume baseline, that is, face upward head Under state, the ratio of open wiring to the vertical distance between nose baseline and nose baseline to the vertical distance between volume baseline is greater than 1:1；When When human face posture is become bowing from positive face, open wiring to the face part between nose baseline is than the face between nose baseline to volume baseline The distance of partial distance taking lens is close, becomes smaller under the relatively positive face-like state of corresponding open wiring to the vertical distance between nose baseline, Become larger under nose baseline to the relatively positive face-like state of the vertical distance between volume baseline, that is, bows under state, open wiring is between nose baseline Vertical distance and nose baseline to the vertical distance between volume baseline ratio be less than 1:1.

Based on above-mentioned analysis, the present invention can be by judging open wiring to the vertical distance and nose baseline to volume between nose baseline Size between the ratio and first threshold Y of vertical distance between baseline judges that the human face posture of face image to be processed is No is to bow or face upward head.In view of the otherness of measurement error and the distribution of different faces face, it is preferred that first threshold Y's takes Value is a numberical range rather than a specific value, exemplary, and first threshold Y is [0.8,1.2].

Refering to what is shown in Fig. 8, the first key point A1 represents vertical position of the open wiring in face image, the second key point A2 generation Vertical position of the table nose baseline in face image, third key point A3 represent vertical position of the volume baseline in face image, Vertical distance Y1 between first key point A1 to the second key point A2 represents open wiring to the vertical distance between nose baseline, and second Vertical distance between key point A2 to third key point A3 represents nose baseline to the vertical distance Y2 between volume baseline.

If ratio of the ratio of Y1 and Y2 equal to first threshold Y, i.e. Y1 and Y2 is interior in [0.8,1.2], then the face image Human face posture be positive face；If ratio of the ratio of Y1 and Y2 greater than first threshold Y, i.e. Y1 and Y2 is greater than 1.2, then the face schemes The human face posture of picture is to face upward head；If ratio of the ratio of Y1 and Y2 less than first threshold Y, i.e. Y1 and Y2 is less than 0.8, then the face The human face posture of image is to bow.

Refering to what is shown in Fig. 7, in positive face posture human face face, the left eye angle of left eye between the left basal part of the ear horizontal distance, The left eye angle of left eye is to the right eye angle of horizontal distance, left eye between the right eye angle of left eye to the level between the left eye angle of right eye Distance, right eye left eye angle to the right eye angle of horizontal distance and right eye between the right eye angle of right eye to the level between auris dextra root Distance accounts for 1/5th of face image integral level distance respectively, i.e., the left eye angle of left eye to the level between the left basal part of the ear away from Left eye angle from, left eye is to the right eye angle of horizontal distance, left eye between the right eye angle of left eye between the left eye angle of right eye Horizontal distance, right eye left eye angle to the right eye angle of horizontal distance and right eye between the right eye angle of right eye between auris dextra root Ratio between horizontal distance between whole is 1:1:1:1:1:1.It finds after study, when the human face posture of face image is from just When face variation is left side of the face or right side face, the left eye angle to the left eye angle of horizontal distance, left eye between the left basal part of the ear a to left side of left eye Left eye of the right eye angle of horizontal distance, left eye between the right eye angle of eye to horizontal distance, right eye between the left eye angle of right eye Angle is to the right eye angle of horizontal distance and right eye between the right eye angle of right eye to the horizontal distance between auris dextra root relative to positive face Different degrees of variation occurs under posture, therefore, we can be both any between above-mentioned five horizontal distances by judging Ratio situation of change judge whether the human face posture of the face image is left side of the face or right side face.

Below in conjunction with the left eye angle to the left eye angle of horizontal distance and right eye between the right eye angle of left eye to the right side of left eye Size relation between the ratio and second threshold X of horizontal distance between the right eye angle of eye, judges the face of image to be processed Whether posture is left side head or right side head.In view of the otherness of measurement error and different faces face, it is preferred that second threshold The value of X is a numberical range rather than a specific value, exemplary, and first threshold X is [0.9,1.1].Certainly, herein It is merely illustrative of, does not represent the embodiment of the present invention and judge whether the human face posture of image to be processed is left side head or right side head Method be confined to this.

Specifically, refering to what is shown in Fig. 8, the 4th key point B1 represents horizontal position of the left eye angle of left eye in face image It sets, the 5th key point B2 represents horizontal position of the right eye angle of left eye in face image, and the 6th key point B3 represents right eye Horizontal position of the left eye angle in face image, the 7th key point B4 represent horizontal position of the right eye angle of right eye in face image It sets, the horizontal distance between the 4th key point B1 to the 5th key point B2 represents the left eye angle of left eye between the right eye angle of left eye Horizontal distance, right eye is arrived at the left eye angle that the horizontal distance X2 between the 6th key point B3 to the 7th key point B4 represents right eye Horizontal distance between right eye angle.

If ratio of the ratio of X1 and X2 equal to second threshold X, i.e. X1 and X2 is interior in [0.9,1.1], then the face image Human face posture be positive face；If ratio of the ratio of X1 and X2 greater than second threshold X, i.e. X1 and X2 is greater than 1.1, then the face schemes The human face posture of picture is left side head；If ratio of the ratio of X1 and X2 less than second threshold X, i.e. X1 and X2 is less than 0.9, then the face The human face posture of portion's image is right side head.

It further, should if the ratio of Y1 and Y2 is greater than first threshold Y and the ratio of X1 and X2 is greater than second threshold X The human face posture of face image is to face upward head and left side head；If the ratio of Y1 and Y2 is greater than first threshold Y and the ratio of X1 and X2 is small In second threshold X, then the human face posture of the face image is to face upward head and right side head；If the ratio of Y1 and Y2 is less than first threshold Y And the ratio of X1 and X2 is greater than second threshold X, then the human face posture of the face image is to bow and left side head, if the ratio of Y1 and Y2 The ratio that value is less than first threshold Y and X1 and X2 is less than second threshold X, then the human face posture of the face image is to bow and right side Head.

Step 110: according to the human face posture of the face image, correcting the corresponding Ellipse Neighborhood radius of the key point.

Specifically, the implementation procedure of step 110 is as follows:

The horizontal axis of the face if the human face posture is positive, the Ellipse Neighborhood is equal to the first preset value, the Ellipse Neighborhood The longitudinal axis is equal to the first preset value.

If the human face posture is to face upward head or bow, the horizontal axis of the Ellipse Neighborhood is equal to the first preset value, the ellipse The longitudinal axis of neighborhood is equal to the product of the first preset value and the first default weight.

If the human face posture is left side head or right side head, the horizontal axis of the Ellipse Neighborhood is equal to the first preset value and second The longitudinal axis of the product of default weight, the Ellipse Neighborhood is equal to the first preset value.

Specifically, refering to what is shown in Fig. 9, the Ellipse Neighborhood horizontal axis of key point O be W, longitudinal axis H.According to the judgement of step 100 As a result, the face if the human face posture of face image to be processed is positive, the Ellipse Neighborhood horizontal axis W of key point O is equal to the first preset value D, longitudinal axis H is equal to the first preset value D；If the human face posture of face image to be processed is to face upward head or bow, the ellipse of key point O Neighborhood horizontal axis W is equal to the product that the first preset value D, longitudinal axis H are equal to the default Quan Chong &1 of the first preset value D and first；If to be processed The human face posture of face image be left side head or right side head, then the Ellipse Neighborhood horizontal axis W of key point O be equal to the first preset value D with The product of second default Quan Chong &2, longitudinal axis H are equal to the first preset value D.

Further, firstly, face image to be processed is divided into four regions, respectively first area, second area, Three regions and the fourth region, wherein first area, second area, third region and the fourth region are distributed along clockwise, and first Region is located on the fourth region vertical direction.

It is exemplary, refering to what is shown in Fig. 10, first area is located at half face of a left side of face and is located at nose baseline area above, second Region is located at half face of the right side of face and is located at nose baseline area above, and third region is located at half face of the right side of face and is located at nose baseline Following region, the fourth region are located at half face of a left side of face and are located at nose baseline following region.

Specifically, according to the human face modeling of the face image to be processed of step 100 as a result, if face image to be processed Human face posture be positive face, then the ellipse for being located at the key point O of first area, second area, third region and the fourth region is adjacent Domain horizontal axis W is equal to the first preset value D, longitudinal axis H and is equal to the first preset value D.

If the human face posture of face image to be processed is to face upward head, it is located at first area, second area, third region and the It is default equal to the first preset value D and first that the Ellipse Neighborhood horizontal axis W of four-range key point O is equal to the first preset value D, longitudinal axis H The product of Quan Chong &1.Wherein, when key point O is located at first area and second area, the first default Quan Chong &1 ascends the throne less than 1 In the longitudinal axis H of first area and the Ellipse Neighborhood of the key point O of second area less than the first preset value D；When key point O is located at the When three regions and the fourth region, the first default weight is greater than 1, that is, is located at the ellipse of the key point O of first area and second area The longitudinal axis H of circular neighborhood is greater than the first preset value D.

If the human face posture of face image to be processed is to bow, it is located at first area, second area, third region and the It is default equal to the first preset value D and first that the Ellipse Neighborhood horizontal axis W of four-range key point O is equal to the first preset value D, longitudinal axis H The product of Quan Chong &1.Wherein, when key point O is located at first area and second area, the first default Quan Chong &1 is greater than 1, ascends the throne It is greater than the first preset value D in the longitudinal axis H of first area and the Ellipse Neighborhood of the key point O of second area；When key point O is located at the When three regions and the fourth region, the first default weight is located at the ellipse of the key point O of first area and second area less than 1 The longitudinal axis H of circular neighborhood is less than the first preset value D.

If the human face posture of face image to be processed be left side head, be located at first area, second area, third region and The Ellipse Neighborhood horizontal axis W of the key point O of the fourth region is equal to product, the longitudinal axis H etc. of the default Quan Chong &2 of the first preset value D and second In the first preset value D.Wherein, when key point O is located at first area and third region, the second default Quan Chong &2 is greater than 1, ascends the throne It is greater than the first preset value D in the horizontal axis W of first area and the Ellipse Neighborhood of the key point O of second area；When key point O is located at the When two regions and the fourth region, the second default weight is located at the ellipse of the key point O of first area and second area less than 1 The horizontal axis W of circular neighborhood is less than the first preset value D.

If the human face posture of face image to be processed be right side head, be located at first area, second area, third region and The Ellipse Neighborhood horizontal axis W of the key point O of the fourth region is equal to product, the longitudinal axis H etc. of the default Quan Chong &2 of the first preset value D and second In the first preset value D.Wherein, when key point O is located at first area and third region, the second default Quan Chong &2 ascends the throne less than 1 In the horizontal axis W of first area and the Ellipse Neighborhood of the key point O of second area less than the first preset value D；When key point O is located at the When two regions and the fourth region, the second default weight is greater than 1, that is, is located at the ellipse of the key point O of first area and second area The horizontal axis W of circular neighborhood is greater than the first preset value D.

If the human face posture of face image to be processed is to face upward head and left side head, it is located at first area, second area, third Product of the Ellipse Neighborhood horizontal axis W of the key point O of region and the fourth region equal to the default Quan Chong &2 of the first preset value D and second, Longitudinal axis H is equal to the product of the default Quan Chong &1 of the first preset value D and first.Wherein, when key point O is located at first area, second Default weight is greater than 1, i.e. the horizontal axis W positioned at the Ellipse Neighborhood of the key point O of first area is greater than the first preset value D, and first Weight is preset less than 1, that is, is located at the longitudinal axis H of the Ellipse Neighborhood of the key point O of first area less than the first preset value D；Work as pass When key point O is located at second area, the second default weight is located at the cross of the Ellipse Neighborhood of the key point O of first area less than 1 For axis W less than the first preset value D, the first default weight is located at the vertical of the Ellipse Neighborhood of the key point O of first area less than 1 Axis H is less than the first preset value D；When key point O is located at third region, the second default weight is greater than 1, that is, is located at first area Key point O Ellipse Neighborhood horizontal axis W be greater than the first preset value D, the first default weight be greater than 1, that is, be located at first area Key point O Ellipse Neighborhood longitudinal axis H be greater than the first preset value D；When key point O is located at the fourth region, the second default power Again less than 1, that is, it is located at the horizontal axis W of the Ellipse Neighborhood of the key point O of first area less than the first preset value D, the first default power It is greater than 1 again, i.e. the longitudinal axis H positioned at the Ellipse Neighborhood of the key point O of first area is greater than the first preset value D.

If the human face posture of face image to be processed is to face upward head and right side head, it is located at first area, second area, third Product of the Ellipse Neighborhood horizontal axis W of the key point O of region and the fourth region equal to the default Quan Chong &2 of the first preset value D and second, Longitudinal axis H is equal to the product of the default Quan Chong &1 of the first preset value D and first.Wherein, when key point O is located at first area, second Weight is preset less than 1, i.e., positioned at first area key point O Ellipse Neighborhood horizontal axis W less than the first preset value D, first Weight is preset less than 1, that is, is located at the longitudinal axis H of the Ellipse Neighborhood of the key point O of first area less than the first preset value D；Work as pass When key point O is located at second area, the second default weight is greater than 1, that is, is located at the cross of the Ellipse Neighborhood of the key point O of first area Axis W is greater than the first preset value D, the first default weight is located at the vertical of the Ellipse Neighborhood of the key point O of first area less than 1 Axis H is less than the first preset value D；When key point O is located at third region, the second default weight is located at first area less than 1 Key point O Ellipse Neighborhood horizontal axis W less than the first preset value D, the first default weight is greater than 1, that is, is located at first area Key point O Ellipse Neighborhood longitudinal axis H be greater than the first preset value D；When key point O is located at the fourth region, the second default power It is greater than 1 again, i.e. the horizontal axis W positioned at the Ellipse Neighborhood of the key point O of first area is greater than the first preset value D, the first default power It is greater than 1 again, i.e. the longitudinal axis H positioned at the Ellipse Neighborhood of the key point O of first area is greater than the first preset value D.

If the human face posture of face image to be processed is to bow and left side head, it is located at first area, second area, third Product of the Ellipse Neighborhood horizontal axis W of the key point O of region and the fourth region equal to the default Quan Chong &2 of the first preset value D and second, Longitudinal axis H is equal to the product of the default Quan Chong &1 of the first preset value D and first.Wherein, when key point O is located at first area, second Default weight is greater than 1, i.e. the horizontal axis W positioned at the Ellipse Neighborhood of the key point O of first area is greater than the first preset value D, and first Default weight is greater than 1, i.e. the longitudinal axis H positioned at the Ellipse Neighborhood of the key point O of first area is greater than the first preset value D；Work as pass When key point O is located at second area, the second default weight is located at the cross of the Ellipse Neighborhood of the key point O of first area less than 1 For axis W less than the first preset value D, the first default weight is greater than 1, that is, is located at the vertical of the Ellipse Neighborhood of the key point O of first area Axis H is greater than the first preset value D；When key point O is located at third region, the second default weight is greater than 1, that is, is located at first area Key point O Ellipse Neighborhood horizontal axis W be greater than the first preset value D, the first default weight less than 1, that is, be located at first area Key point O Ellipse Neighborhood longitudinal axis H less than the first preset value D；When key point O is located at the fourth region, the second default power Again less than 1, that is, it is located at the horizontal axis W of the Ellipse Neighborhood of the key point O of first area less than the first preset value D, the first default power Again less than 1, that is, it is located at the longitudinal axis H of the Ellipse Neighborhood of the key point O of first area less than the first preset value D.

If the human face posture of face image to be processed is to face upward to bow and right side head, it is located at first area, second area, the The Ellipse Neighborhood horizontal axis W of the key point O of three regions and the fourth region multiplies equal to the first preset value D and the second default Quan Chong &2's Product, longitudinal axis H are equal to the product of the default Quan Chong &1 of the first preset value D and first.Wherein, when key point O is located at first area, the Two default weights less than 1, i.e., positioned at first area key point O Ellipse Neighborhood horizontal axis W less than the first preset value D, the One default weight is greater than 1, i.e. the longitudinal axis H positioned at the Ellipse Neighborhood of the key point O of first area is greater than the first preset value D；When When key point O is located at second area, the second default weight is greater than 1, that is, is located at the Ellipse Neighborhood of the key point O of first area Horizontal axis W is greater than the first preset value D, and the first default weight is greater than 1, that is, is located at the Ellipse Neighborhood of the key point O of first area Longitudinal axis H is greater than the first preset value D；When key point O is located at third region, the second default weight is located at the firstth area less than 1 For the horizontal axis W of the Ellipse Neighborhood of the key point O in domain less than the first preset value D, the first default weight is located at the firstth area less than 1 The longitudinal axis H of the Ellipse Neighborhood of the key point O in domain is less than the first preset value D；When key point O is located at the fourth region, second is default Weight is greater than 1, i.e. the horizontal axis W positioned at the Ellipse Neighborhood of the key point O of first area is greater than the first preset value D, and first is default Weight is located at the longitudinal axis H of the Ellipse Neighborhood of the key point O of first area less than the first preset value D less than 1.

It is exemplary, , &1=1-esin α when the first default Quan Chong &1 is less than 1, wherein α is people in face image to be processed The side brilliance degree of face posture；, &1=1+esin α when first default Quan Chong &1 is greater than 1, wherein α is in face image to be processed The side brilliance degree of human face posture；When second default Quan Chong &2 is less than 1, Shi Li &2=0.8；When second default Quan Chong &2 is greater than 1, Shi Li &2=1.2.

It is exemplary, the face image of same people, Ellipse Neighborhood horizontal axis W, longitudinal axis H and the human face posture and key of key point O Corresponding relationship between the position of point O is as shown in following table table one:

Table one

Step 120: LBP feature being carried out to the key point using the Ellipse Neighborhood radius to the face image and is mentioned It takes.

Specifically, needing to scheme face to be processed before carrying out LBP feature extraction to the key point of face image to be processed As being normalized, by the size scaling of face image to be processed to same size.It is exemplary, if face to be processed schemes Actual range is L between the double eyeball of face as in, if the human face posture of face image to be processed is not left side head or right side head, Face image to be processed is zoomed in and out until wherein distance is L the double eyeball of face；If the face of face image to be processed Posture is left side head or right side head, then zooms in and out between face image to be processed until wherein distance is the double eyeball of face Lcos α, wherein α is the side brilliance degree of human face posture in face image to be processed.

Further, centered on each key point, one piece of region in face image to be processed is taken, and by the region It is divided into the identical image block of several sizes, LBP feature extraction then is carried out to each pixel of each image block.Right During each pixel carries out LBP feature extraction, round LBP operator is used, wherein each pixel neighborhood of a point half Diameter uses the numerical value of the horizontal axis W and longitudinal axis H of the Ellipse Neighborhood determined in step 110 according to human face posture and key point position, into And the pixel value of several sampled points Ellipse Neighborhood Nei is obtained, it will be in the pixel value of several sampled points and the Ellipse Neighborhood The pixel value of heart point is compared, and the sampling point position greater than the central point pixel value is labeled as 1, is otherwise labeled as 0.Then, According to the label of each sampled point as a result, the binary number of certain digit is obtained in a certain order, then according to mould of equal value The binary number is converted decimal number by formula, and the histogram of the image block is generated according to the decimal number, finally by each figure As the histogram of block is connected, the LBP feature vector of the key point is generated.

Further, about the LBP feature extraction algorithm of key point, the present invention is not described in detail, those skilled in the art Member can refer to the prior art.

LBP feature extracting method provided in an embodiment of the present invention based on face's key point, first according to face image Key point determines the human face posture of the face image, and then corrects different face's key points according to the human face posture of the face image Corresponding Ellipse Neighborhood radius establishes the human face posture of Ellipse Neighborhood radius and the face image and the position of the key point Corresponding relationship between setting, and then the revised Ellipse Neighborhood radius of use extracts the LBP feature of face's key point, to keep away Exempt under same person's different faces posture, face image local scale ratio is different, and the LBP for causing same key point to be extracted is special The larger problem of difference is levied, improves the validity of the LBP feature extraction under different faces posture based on face's key point, in turn Improve face recognition accuracy rate.

Embodiment two

Figure 11 shows another LBP feature extraction process signal based on face's key point provided in an embodiment of the present invention Figure, as shown in figure 11, being somebody's turn to do the LBP characteristic extraction procedure based on face's key point includes:

Step 200: the key point based on face image to be processed estimates the human face posture of the face image.

Specifically, the implementation procedure of step 200 is identical as the step 100 of above-described embodiment one, before be discussed in detail, The present invention does not do to tire out herein and state, and specifically please refers to the step 100 of embodiment one.

Step 210: determining that the key point is corresponding in preset mapping table according to the human face posture of the face image Ellipse Neighborhood radius.

Specifically, searching face to be processed according to the human face posture of the face image to be processed determined in above-mentioned steps 200 The key point of image Ellipse Neighborhood radius corresponding in preset mapping table, wherein the preset mapping table is same for characterizing Personal face image, the key point in the different face areas under different faces posture carry out in LBP characteristic extraction procedure Corresponding Ellipse Neighborhood radius size.It is exemplary, refering to what is shown in Fig. 10, face is divided into four regions, respectively first area, Second area, third region and the fourth region, wherein first area, second area, third region and the fourth region are along clockwise Distribution, and first area is located on the fourth region vertical direction, wherein first area is located at half face of a left side of face and is located at Nose baseline area above, second area are located at half face of the right side of face and are located at nose baseline area above, and third region is located at face Half face of the right side and be located at nose baseline following region, the fourth region is located at half face of a left side of face and positioned at nose baseline following region.

Further, in the preset mapping table, the face appearance of face image key point O to be processed and face image to be processed The corresponding relationship of state and key point O between the distributing position on face image to be processed is as follows:

The face if human face posture of face image to be processed is positive is located at first area, second area, third region and the The Ellipse Neighborhood horizontal axis W of four-range key point O is equal to the first preset value D, longitudinal axis H and is equal to the first preset value D.

It is exemplary, , &1=1-esin α when the first default Quan Chong &1 is less than 1, wherein α is people in face image to be processed The side brilliance degree of face posture；, &1=1+esin α when first default Quan Chong &1 is greater than 1, wherein α is in face image to be processed The side brilliance degree of human face posture；When second default Quan Chong &2 is less than 1, Shi Li &2=0.7；When second default Quan Chong &2 is greater than 1, Shi Li &2=1.1.

It is exemplary, the face image of same people, Ellipse Neighborhood horizontal axis W, longitudinal axis H and the human face posture and key of key point O Corresponding relationship between the position of point O is as shown in following table table two, i.e., the preset mapping table is as shown in following table table two:

Table two

Step 220: LBP feature being carried out to the key point using the Ellipse Neighborhood radius to the face image and is mentioned It takes.

Specifically, the implementation procedure of step 220 is identical as the step 120 of above-described embodiment one, before be discussed in detail, The present invention does not do to tire out herein and state, and specifically please refers to the step 120 of embodiment one.

LBP feature extracting method provided in an embodiment of the present invention based on face's key point, first according to face image Key point determines the human face posture of the face image, and then according to the human face posture of the face image, searches face's figure to be processed The key point of picture Ellipse Neighborhood radius corresponding in preset mapping table, wherein the preset mapping table is same for characterizing The face image of people, it is right in LBP characteristic extraction procedure that the key point in the different face areas under different faces posture carries out The Ellipse Neighborhood radius size answered.The preset mapping table of the embodiment of the present invention establishes Ellipse Neighborhood radius and the face image Corresponding relationship between human face posture and the position of the key point, and then mentioned using the Ellipse Neighborhood radius in preset mapping table The LBP feature of face's key point is taken, so as to avoid under same person's different faces posture, face image local scale ratio Difference, the larger problem of the LBP characteristic difference for causing same key point to be extracted are improved and are closed under different faces posture based on face The validity of the LBP feature extraction of key point, and then improve face recognition accuracy rate.

The LBP feature extracting method based on face's key point based on the above embodiment, the embodiment of the present invention also provide one LBP feature deriving means of the kind based on face's key point, are carried out using LBP feature extracting method provided by the above embodiment wait locate The LBP feature extraction of face image is managed, the device is as shown in figure 12, comprising:

Critical point detection module 301 determines LBP feature to be extracted for detecting the key point of face image to be processed Face's key point position.

Image processing module 302 determines the face according to the distance for obtaining the distance between described key point The human face posture of image, and the radius that the key point corresponds to Ellipse Neighborhood is corrected according to the human face posture.

Characteristic extracting module 303, for using the radius of the Ellipse Neighborhood to the key point face image Carry out LBP feature extraction.

Specifically, critical point detection module 301 is used to detect the key point of face image to be processed, face to be processed is obtained Face's key point position of LBP feature to be extracted in image.Further, critical point detection module 301 is for detecting generation respectively The position of facial contour, the position of eyebrow, the position of eyes, the position of nose and the position of mouth in table face image to be processed Face face exterior feature key point, eyebrow key point, eyes key point, nose key point and mouth key point etc., and then based on key The position of point, is divided into first area, second area, third region and the fourth region for face image to be processed with reference to shown in Figure 10 Four regions, wherein first area, second area, third region and the fourth region are along distribution clockwise, and first area is located at Exemplary on the fourth region vertical direction, first area is located at half face of a left side of face and is located at nose baseline area above, the Two regions are located at half face of the right side of face and are located at nose baseline area above, and third region is located at half face of the right side of face and is located at nose bottom Line following region, the fourth region are located at half face of a left side of face and are located at nose baseline following region.

Specifically, image processing module 302 is used to obtain the distance between key point to be processed, and according to the face got The distance between portion's key point determines the human face posture for just handling face image, finally according to determining face image to be processed Human face posture, to face image to be processed carry out scaling, and determine different faces posture under different face areas in Key point carry out LBP characteristic extraction procedure in corresponding Ellipse Neighborhood radius size.Specifically, image processing module 302 According to the distance between the face's key point got, determines and just handle the human face posture of face image and according to determining wait locate The human face posture for managing face image carries out scaling to face image to be processed, and determines the difference under different faces posture The method that key point in face area carries out corresponding Ellipse Neighborhood radius size in LBP characteristic extraction procedure, with front reality It is identical with the method that describes in step 210 with the step 200 of 110 and embodiment two to apply the step 100 of example one, it is of the invention herein Do not do tired state.

Specifically, characteristic extracting module 303 is used to determine the face image after scaling using in image processing module 302 The Ellipse Neighborhood radius of face's key point LBP feature extraction, specific LBP feature extracting method and mistake are carried out to the key point Journey, please refers to the step 120 of embodiment one and the step 220 of embodiment two, and the present invention does not do tired state herein.

LBP feature deriving means provided in an embodiment of the present invention based on face's key point, first according to face image Key point determines the human face posture of the face image, and then according to the human face posture of the face image, determines face's figure to be processed As the key point in the different face areas under the human face posture carries out corresponding Ellipse Neighborhood half in LBP characteristic extraction procedure Diameter size.The LBP feature deriving means of the embodiment of the present invention establish the human face posture of Ellipse Neighborhood radius Yu the face image And the corresponding relationship between the position of the key point, and then the LBP of face's key point is extracted using the Ellipse Neighborhood radius Feature, so as to avoid under same person's different faces posture, face image local scale ratio is different, leads to same key point The larger problem of the LBP characteristic difference of extraction, improves the LBP feature extraction under different faces posture based on face's key point Validity, and then improve face recognition accuracy rate.

The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs The processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices is instructed, so that A stream in flow chart can be achieved by the instruction that the computer or the processor of other programmable data processing devices execute The function of being specified in journey or multiple processes and/or one or more blocks of the block diagram.

These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates, Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or The function of being specified in multiple boxes.

These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or The instruction executed on other programmable devices is provided for realizing in one process or multiple processes and/or block diagrams of flow chart One box or multiple boxes in specify function the step of.

Although preferred embodiments of the present invention have been described, it is created once a person skilled in the art knows basic Property concept, then additional changes and modifications can be made to these embodiments.So it includes excellent that the following claims are intended to be interpreted as It selects embodiment and falls into all change and modification of the scope of the invention.

Obviously, those skilled in the art can carry out various modification and variations without departing from this hair to the embodiment of the present invention The spirit and scope of bright embodiment.In this way, if these modifications and variations of the embodiment of the present invention belong to the claims in the present invention And its within the scope of equivalent technologies, then the present invention is also intended to include these modifications and variations.

Claims

1. a kind of LBP feature extracting method based on face's key point characterized by comprising

2. a kind of LBP feature extracting method based on face's key point characterized by comprising

The key point corresponding Ellipse Neighborhood radius in preset mapping table is determined according to the human face posture of the face image；

3. method according to claim 1 or 2, which is characterized in that the key point based on face image to be processed is estimated Count the human face posture of the face image, comprising:

Obtain the first vertical distance and the second vertical distance on the face image vertical direction, wherein described first is vertical Distance and second vertical distance are distributed from up to down along the vertical direction；

Ratio between first vertical distance and second vertical distance is greater than first threshold, then the face image Human face posture is to bow；Or

Ratio between first vertical distance and second vertical distance is less than first threshold, then the face image Human face posture is to face upward head.

4. method according to claim 1 or claim 2, which is characterized in that the key point based on face image to be processed, estimation The human face posture of the face image, further includes:

Obtain the first level distance and the second horizontal distance in the face image horizontal direction, wherein the first level Distance and second horizontal distance are distributed from left to right in the horizontal direction；

Ratio between the first level distance and second horizontal distance is greater than second threshold, then the face image Human face posture is right side head；Or

Ratio between the first level distance and second horizontal distance is less than second threshold, then the face image Human face posture is left side head.

5. according to claim 1 or any one of 2 the methods, which is characterized in that the face appearance according to the face image State corrects the corresponding Ellipse Neighborhood radius of the key point, comprising:

The horizontal axis of the face if the human face posture is positive, the Ellipse Neighborhood is equal to the first preset value, the longitudinal axis of the Ellipse Neighborhood Equal to the first preset value；Or

If the human face posture is to face upward head or bow, the horizontal axis of the Ellipse Neighborhood is equal to the first preset value, the Ellipse Neighborhood The longitudinal axis be equal to the first preset value and the first default weight product；Or

If the human face posture is left side head or right side head, the horizontal axis of the Ellipse Neighborhood, which is equal to the first preset value and second, to be preset The longitudinal axis of the product of weight, the Ellipse Neighborhood is equal to the first preset value.

6. method according to claim 5, which is characterized in that the face image includes first area, second area, third Region and the fourth region, wherein the first area, the second area, the third region and the fourth region are along suitable Hour hands distribution, and the first area is located on the fourth region vertical direction.

7. method according to claim 6, which is characterized in that the human face posture according to the face image corrects institute State the corresponding Ellipse Neighborhood radius of key point, comprising:

The face if human face posture is positive, the first area, the second area, the third region and the fourth region The horizontal axis of the Ellipse Neighborhood and the longitudinal axis of the Ellipse Neighborhood be equal to the first preset value.

8. method according to claim 6, which is characterized in that the human face posture according to the face image corrects institute State the corresponding Ellipse Neighborhood radius of key point, further includes:

If the human face posture is to face upward head, the first area, the second area, the third region and the fourth region The horizontal axis be equal to first preset value, the longitudinal axis of the first area and the second area is default less than first The longitudinal axis of value, the third region and the fourth region is greater than first preset value.

9. method according to claim 6, which is characterized in that the human face posture according to the face image corrects institute State the corresponding Ellipse Neighborhood radius of key point, further includes:

If the human face posture is to bow, the first area, the second area, the third region and the fourth region The horizontal axis be equal to first preset value, it is default that the longitudinal axis of the first area and the second area is greater than first The longitudinal axis of value, the third region and the fourth region is less than first preset value.

10. method according to claim 6, which is characterized in that the human face posture according to the face image corrects institute State the corresponding Ellipse Neighborhood radius of key point, further includes:

If the human face posture is left side head, it is default that the horizontal axis of the first area and the fourth region is greater than first Value, the horizontal axis in the second area and the third region are less than first preset value, the first area, described the The longitudinal axis in two regions, the third region and the fourth region is equal to first preset value.

11. method according to claim 6, which is characterized in that the human face posture according to the face image corrects institute State the corresponding Ellipse Neighborhood radius of key point, further includes:

If the human face posture is right side head, the horizontal axis of the first area and the fourth region is default less than first Value, the horizontal axis in the second area and the third region are greater than first preset value, the first area, described the The longitudinal axis in two regions, the third region and the fourth region is equal to first preset value.

12. method according to claim 6, which is characterized in that the human face posture according to the face image corrects institute State the corresponding Ellipse Neighborhood radius of key point, further includes:

If the human face posture is to face upward head and left side head, the horizontal axis of the first area and the fourth region is greater than described The horizontal axis in the first preset value, the second area and the third region is less than first preset value, firstth area The longitudinal axis of domain and the second area is less than the first preset value, the longitudinal axis in the third region and the fourth region Greater than the first preset value.

13. method according to claim 6, which is characterized in that the human face posture according to the face image corrects institute State the corresponding Ellipse Neighborhood radius of key point, further includes:

If the human face posture is to face upward head and right side head, the horizontal axis of the first area and the fourth region is less than described The horizontal axis in the first preset value, the second area and the third region is greater than first preset value, firstth area The longitudinal axis of domain and the second area is greater than the first preset value, the longitudinal axis in the third region and the fourth region Less than the first preset value.

14. method according to claim 6, which is characterized in that the human face posture according to the face image corrects institute State the corresponding Ellipse Neighborhood radius of key point, further includes:

If the human face posture is to bow and left side head, the horizontal axis of the first area and the fourth region is greater than described The horizontal axis in the first preset value, the second area and the third region is less than first preset value, firstth area The longitudinal axis of domain and the second area is greater than the first preset value, the longitudinal axis in the third region and the fourth region Less than the first preset value.

15. method according to claim 6, which is characterized in that the human face posture according to the face image corrects institute State the corresponding Ellipse Neighborhood radius of key point, further includes:

If the human face posture is to face upward head and left side head, the horizontal axis of the first area and the fourth region is less than described The horizontal axis in the first preset value, the second area and the third region is greater than first preset value, firstth area The longitudinal axis of domain and the second area is less than the first preset value, the longitudinal axis in the third region and the fourth region Greater than the first preset value.

16. method according to claim 6, which is characterized in that preset mapping table is used to characterize under different faces posture, is different The corresponding Ellipse Neighborhood radius of the key point in face area, in which:

The face if human face posture is positive, the first area, the second area, the third region and the fourth region The horizontal axis of the Ellipse Neighborhood and the longitudinal axis of the Ellipse Neighborhood be equal to the first preset value；

If the human face posture is to face upward head, the first area, the second area, the third region and the fourth region The horizontal axis be equal to first preset value, the longitudinal axis of the first area and the second area is default less than first The longitudinal axis of value, the third region and the fourth region is greater than first preset value；

If the human face posture is to bow, the first area, the second area, the third region and the fourth region The horizontal axis be equal to first preset value, it is default that the longitudinal axis of the first area and the second area is greater than first The longitudinal axis of value, the third region and the fourth region is less than first preset value；

If the human face posture is left side head, it is default that the horizontal axis of the first area and the fourth region is greater than first Value, the horizontal axis in the second area and the third region are less than first preset value, the first area, described the The longitudinal axis in two regions, the third region and the fourth region is equal to first preset value；

If the human face posture is right side head, the horizontal axis of the first area and the fourth region is default less than first Value, the horizontal axis in the second area and the third region are greater than first preset value, the first area, described the The longitudinal axis in two regions, the third region and the fourth region is equal to first preset value；

If the human face posture is to face upward head and left side head, the horizontal axis of the first area and the fourth region is greater than described The horizontal axis in the first preset value, the second area and the third region is less than first preset value, firstth area The longitudinal axis of domain and the second area is less than the first preset value, the longitudinal axis in the third region and the fourth region Greater than the first preset value；

If the human face posture is to face upward head and right side head, the horizontal axis of the first area and the fourth region is less than described The horizontal axis in the first preset value, the second area and the third region is greater than first preset value, firstth area The longitudinal axis of domain and the second area is greater than the first preset value, the longitudinal axis in the third region and the fourth region Less than the first preset value；

If the human face posture is to bow and left side head, the horizontal axis of the first area and the fourth region is greater than described The horizontal axis in the first preset value, the second area and the third region is less than first preset value, firstth area The longitudinal axis of domain and the second area is greater than the first preset value, the longitudinal axis in the third region and the fourth region Less than the first preset value；

17. a kind of LBP feature deriving means based on face's key point characterized by comprising

Image processing module determines the face image according to the distance for obtaining the distance between described key point Human face posture, and the radius that the key point corresponds to Ellipse Neighborhood is corrected according to the human face posture；

Characteristic extracting module, for carrying out LBP to the key point using the radius of the Ellipse Neighborhood to the face image Feature extraction.

18. 7 described device according to claim 1, which is characterized in that described image processing module is specifically used for:

Obtain the first vertical distance and the second vertical distance on the face image vertical direction, wherein described first is vertical Distance and second vertical distance are distributed from up to down along the vertical direction, first vertical distance with described second vertically away from Ratio between is greater than first threshold, then the human face posture of the face image be bow or first vertical distance with Ratio between second vertical distance is less than first threshold, then the human face posture of the face image is to face upward head；

Obtain the first level distance and the second horizontal distance in the face image horizontal direction, wherein the first level Distance and second horizontal distance are distributed from left to right in the horizontal direction, the first level distance with described second it is horizontal away from Ratio between is greater than second threshold, then the human face posture of the face image is right side head or the first level distance Ratio between second horizontal distance is less than second threshold, then the human face posture of the face image is left side head.

19. 7 described device according to claim 1, which is characterized in that described image processing module is specifically used for:

The face if human face posture is positive determines the described oval adjacent of first area, second area, third region and the fourth region The longitudinal axis of the horizontal axis in domain and the Ellipse Neighborhood is equal to the first preset value；Or

If the human face posture is to face upward head, the first area, the second area, the third region and the described 4th are determined The horizontal axis in region is equal to first preset value, and the longitudinal axis of the first area and the second area is less than first The longitudinal axis of preset value, the third region and the fourth region is greater than first preset value；Or

If the human face posture is to bow, the first area, the second area, the third region and the described 4th are determined The horizontal axis in region is equal to first preset value, and the longitudinal axis of the first area and the second area is greater than first The longitudinal axis of preset value, the third region and the fourth region is less than first preset value；Or

If the human face posture is left side head, determine that the horizontal axis of the first area and the fourth region is greater than first in advance If value, the horizontal axis in the second area and the third region is less than first preset value, the first area, described The longitudinal axis of second area, the third region and the fourth region is equal to first preset value；Or

If the human face posture is right side head, determine that the horizontal axis of the first area and the fourth region is pre- less than first If value, the horizontal axis in the second area and the third region is greater than first preset value, the first area, described The longitudinal axis of second area, the third region and the fourth region is equal to first preset value；Or

If the human face posture is to face upward head and left side head, determine that the horizontal axis of the first area and the fourth region is greater than The horizontal axis in first preset value, the second area and the third region is less than first preset value, and described the The longitudinal axis of one region and the second area less than the first preset value, the third region and the fourth region it is described The longitudinal axis is greater than the first preset value；Or

If the human face posture is to face upward head and right side head, determine that the horizontal axis of the first area and the fourth region is less than The horizontal axis in first preset value, the second area and the third region is greater than first preset value, and described the The longitudinal axis of one region and the second area is greater than the first preset value, the third region and the fourth region it is described The longitudinal axis is less than the first preset value；Or

If the human face posture is to bow and left side head, determine that the horizontal axis of the first area and the fourth region is greater than The horizontal axis in first preset value, the second area and the third region is less than first preset value, and described the The longitudinal axis of one region and the second area is greater than the first preset value, the third region and the fourth region it is described The longitudinal axis is less than the first preset value；Or

If the human face posture is to face upward head and left side head, determine that the horizontal axis of the first area and the fourth region is less than The horizontal axis in first preset value, the second area and the third region is greater than first preset value, and described the The longitudinal axis of one region and the second area less than the first preset value, the third region and the fourth region it is described The longitudinal axis is greater than the first preset value.