CN109359618A

CN109359618A - A kind of image processing method and its device, equipment and storage medium

Info

Publication number: CN109359618A
Application number: CN201811281028.1A
Authority: CN
Inventors: 黄明杨; 付万增; 石建萍; 曲艺
Original assignee: Beijing Sensetime Technology Development Co Ltd
Current assignee: Beijing Sensetime Technology Development Co Ltd
Priority date: 2018-10-30
Filing date: 2018-10-30
Publication date: 2019-02-19
Anticipated expiration: 2038-10-30
Also published as: CN109359618B

Abstract

The embodiment of the present invention provides a kind of image processing method and its device, equipment and storage medium, wherein the described method includes: obtaining Q key point in facial image, wherein Q is the integer greater than 1；Default key point to be processed is filtered out from the Q key point；Target area to be processed in facial image is determined according to the default key point to be processed, includes the forehead region of facial image in the target area；The target area is divided into N number of subregion, N is the integer more than or equal to 2；Stretching is carried out to the pixel in each subregion respectively, the image that obtains that treated.

Description

A kind of image processing method and its device, equipment and storage medium

Technical field

The present invention relates to technical field of image processing, a kind of image processing method and its device, equipment are related to, but are not limited to And storage medium.

Background technique

With the popularization and application of the development of mobile communication technology, especially mobile network and intelligent terminal, mobile Internet Field development is swift and violent.For at present, the function of intelligent terminal is not simple already for meeting connecting each other for people, and is become Highly important mobile personal recreation terminal in people's daily life.

Nowadays camera function has become the essential function of intelligent terminal, and people can record excellent wink anywhere or anytime Between.And current intelligent terminal all includes front camera mostly, and user is facilitated to carry out self-timer.Liking to be beautiful is the nature of people, greatly It is more good-looking that family all wants the photo oneself taken, therefore has just expedited the emergence of various U.S. face methods.One good face phase, forehead are wanted Full, circumference, bright profit.Traditional full algorithm of 2D forehead is mainly by means of human face detection tech and simple deformation algorithm to figure The forehead of people carries out " filling " operation in piece, to reach face picture " forehead is full " effect.

But there are also significant limitations for the full algorithm of 2D forehead traditional at present.For example, the effect of deformation algorithm very according to Rely the accuracy in human face detection tech, subtle deviation this may result in " shaping failure "；High-precision, key point is intensive Face datection model has high time-consuming, and it is unacceptable with live preview function institute to be that U.S. face camera is taken pictures；In addition, people The three-dimensional multiplicity of forehead, traditional algorithm often only carry out forehead profile to be simply filled with operation, it is difficult to handle different angle, size and The forehead of shape；And 2D U.S. face is difficult to make the face deformation of " three-dimensional sense ", simple deformation can only simply stretch with Forehead profile is pushed, three-dimensional full effect is not achieved.

Summary of the invention

In view of this, an embodiment of the present invention is intended to provide a kind of image processing method and its device, equipment and storage medium, It solves in the full algorithm of forehead in prior art that accuracy rate is low, be easy to cause U.S. face failure and treatment effeciency is low Technical problem can not only be fitted various three-dimensional foreheads, achieve the effect that three-dimensional U.S. face, so that treated image is more Add naturally, and also simplifying implementation to greatly improve treatment effeciency.

The technical solution of the embodiment of the present invention is achieved in that

The embodiment of the present invention provides a kind of image processing method, which comprises

Obtain Q key point in facial image, wherein Q is the integer greater than 1；

Default key point to be processed is filtered out from the Q key point；

Target area to be processed in the facial image is determined according to the default key point to be processed, wherein It include the forehead region of facial image in the target area；

The target area is divided into N number of subregion, wherein N is the integer greater than 2；

Stretching is carried out to the pixel in each subregion respectively, the image that obtains that treated.

In embodiments of the present invention, the default key point to be processed includes forehead region in the facial image Fisrt feature point set accordingly determines target area to be processed in facial image according to the default key point to be processed Domain, comprising:

Obtain the facial angle information and adjusting parameter of the facial image；

According to the fisrt feature point set, facial angle information and adjusting parameter determine the forehead region central point, The first point set on original contour and the second point set on objective contour；

According to target area described in the central point, the second point set and preset second adjustment ratio-dependent.

In embodiments of the present invention, described to be determined according to the fisrt feature point set, facial angle information and adjusting parameter The second point set on the first point set and objective contour on the central point in the forehead region, original contour, comprising:

The central point of the target area is determined according to the fisrt feature point set and the facial angle information；

Second feature point set is determined according to the central point, the fisrt feature point set and adjusting parameter；

Interpolation, phase are carried out according to preset interpolation algorithm respectively to the fisrt feature point set and the second feature point set It should obtain first point set and the second point set.

In embodiments of the present invention, described to determine according to the central point, the fisrt feature point set and adjusting parameter Two feature point sets, comprising:

Determine the first distance between the central point and fisrt feature point；

The first adjustment distance is determined according to the first distance and the adjusting parameter；

The point that the point of fisrt feature described in the distance along the first adjustment direction is the first adjustment distance is determined as accordingly Second feature point, wherein centered on the first adjustment direction point arrive fisrt feature point direction；

It obtains the fisrt feature and concentrates the corresponding second feature point of each fisrt feature point, obtain second feature point Collection.

In embodiments of the present invention, it is described to the fisrt feature point set and the second feature point set respectively according to default Interpolation algorithm carry out interpolation, accordingly obtain first point set and the second point set, comprising:

It is first by the first subset of the fisrt feature point set and the third subset division of corresponding second feature point set The second subset of the fisrt feature point set and the 4th subset division of corresponding second feature point set are the by target subregion Two target subregions；

First subset of the first object subregion and the second subset are calculated according to preset interpolation respectively Method carries out interpolation, obtains corresponding 5th subset and the 6th subset；

The intersection of 5th subset and the 6th subset is determined as the first point set；

The third subset and the 4th subset to the second target subregion are calculated according to preset interpolation respectively Method carries out interpolation, obtains corresponding 7th subset and the 8th subset；

The intersection of 7th subset and the 8th subset is determined as second subset.

In embodiments of the present invention, described according to the central point, the second point set and preset second adjustment ratio-dependent The target area, comprising:

Determine the second distance between the central point and i-th of second point of second point concentration, wherein i=1, 2 ..., N+1, (N+1) are the sum of second point；

According to the second distance and the second adjustment ratio-dependent second adjustment distance；

It is that the point of the second adjustment distance is determined as i-th by i-th of second point described in the distance along second adjustment direction Adjust point, wherein the second adjustment direction is the central point to the direction of i-th of second point；

The target area is determined according to (N+1) a adjustment point.

It is in embodiments of the present invention, described that the target area is divided into N number of subregion, comprising:

The central point, the i-th adjustment point and i+1 adjustment point are sequentially connected and constitute i-th of subregion, wherein i=1, 2 ..., N, (N+1) are the sum of second point.

In embodiments of the present invention, described that stretching is carried out to the pixel in each subregion respectively, it obtains Treated image, comprising:

Obtain the location information of j-th of pixel in i-th of subregion；

According to the location information of j-th of pixel, central point, it is described first point concentration i-th first point, i-th+ 1 first point, i-th of second point of second point concentration, i+1 second point, the i-th adjustment point and i+1 adjustment Point determines stretching function；

J-th of target position is determined according to the location information of j-th of pixel and the stretching function；

The target pixel value of j-th of pixel is determined according to j-th of target position；

The pixel value of j-th of pixel is updated to the target pixel value, the image that obtains that treated.

In embodiments of the present invention, it is described according to the location information of j-th of pixel, it is central point, first point described First point of i-th concentrated, i+1 first point, i-th of second point of second point concentration, the i+1 second Point, the i-th adjustment point and i+1 adjustment point determine stretching function, comprising:

By the 4th line of the central point and j-th of pixel along the central point to j-th of pixel Direction extended, with described i-th first point, the i+1 first point of line intersect at the first intersection point, with it is described I-th of second point, the i+1 second point line intersect at the second intersection point, with the 5th line intersect at third friendship Point, wherein the 5th line is the line of the i-th adjustment point with i+1 adjustment point；

Stretching function is determined according to the 4th distance, the 5th distance and the 6th distance, wherein the 4th distance is institute The distance between central point and first intersection point are stated, the 5th distance is between the central point and second intersection point Distance, the 6th distance is the distance between the central point and the third intersection point.

It is in embodiments of the present invention, described that stretching function is determined according to the 4th distance, the 5th distance and the 6th distance, Include:

Determine the 4th distance and it is described 6th distance between the first ratio, the 5th distance with the described 6th away from The second ratio between；

The first coordinate is determined according to first ratio and second ratio；

The linear equation of first coordinate and the line of origin is determined as the first piecewise function；

The linear equation of first coordinate and the line of preset second coordinate is determined as the second piecewise function；

Stretching function is determined according to first piecewise function and second piecewise function.

In embodiments of the present invention, the location information and the stretching function according to j-th of pixel Determine j-th of target position, comprising:

Between j-th of pixel and the central point is determined according to the location information of j-th of pixel Seven distances；

Determine the third ratio between the 7th distance and the 6th distance；

Using the third ratio as the input of the stretching function, output valve is calculated；

The 8th distance is determined according to the output valve and the 6th distance, wherein the 8th distance is j-th of target The distance between position and the central point；

J-th of mesh is determined according to the location information of the 8th distance, the central point and j-th of pixel Cursor position.

In embodiments of the present invention, the target picture that j-th of pixel is determined according to j-th of target position Element value, comprising:

Coordinate value in response to the target position is integer, and the pixel value of the target position is determined as the jth The target pixel value of a pixel；

Coordinate value in response to the target position is not integer, determines that the target position is corresponding according to preset algorithm Pixel value；

The corresponding pixel value in the target position is determined as to the target pixel value of j-th of pixel.

The embodiment of the present invention provides a kind of image processing apparatus, and described image processing unit includes: the first acquisition module, sieve Modeling block, the first determining module, division module and stretching module, in which:

Described first obtains module, for obtaining Q key point in facial image, wherein Q is the integer greater than 1；

The screening module, for filtering out default key point to be processed from the Q key point；

First determining module, for according to the default key point to be processed determine in the facial image to The target area of processing, wherein include the forehead region of facial image in the target area；

The division module, for the target area to be divided into N number of subregion, wherein N is the integer greater than 2；

The stretching module is obtained for carrying out stretching to the pixel in each subregion respectively Treated image.

In embodiments of the present invention, the default key point to be processed includes forehead region in the facial image Fisrt feature point set, accordingly, the first determining module includes:

First acquisition unit, for obtaining fisrt feature point set, the facial angle letter in forehead region in the facial image Breath and adjusting parameter；

First determination unit, for according to the determination of the fisrt feature point set, facial angle information and adjusting parameter The second point set on the first point set and objective contour on the central point in forehead region, original contour；

Second determination unit, for according to the central point, the second point set and preset second adjustment ratio-dependent Target area.

In embodiments of the present invention, first determination unit, comprising:

First determines subelement, for determining the target according to the fisrt feature point set and the facial angle information The central point in region；

Second determines subelement, for determining second according to the central point, the fisrt feature point set and adjusting parameter Feature point set；

Interpolation subelement is used for the fisrt feature point set and the second feature point set respectively according to preset interpolation Algorithm carries out interpolation, accordingly obtains first point set and the second point set.

In embodiments of the present invention, described second determine that subelement is also used to:

Determine the first distance between the central point and each fisrt feature point；

In embodiments of the present invention, the interpolation subelement is also used to:

In embodiments of the present invention, second determination unit, comprising:

Third determines subelement, for determining between the central point and i-th of second point of second point concentration Second distance, wherein i=1,2 ..., N+1, (N+1) are the sum of second point；

4th determines subelement, for according to the second distance and the second adjustment ratio-dependent second adjustment away from From；

Adjust subelement, for will along second adjustment direction i-th second point described in distance be the second adjustment away from From point be determined as the i-th adjustment point, wherein the second adjustment direction is the central point to the side of i-th of second point To；

5th determines subelement, determines the target area for according to (N+1) a adjustment point.

In embodiments of the present invention, the division module, comprising:

Division unit, for the central point, the i-th adjustment point and i+1 adjustment point to be sequentially connected and constitute i-th of sub-district Domain, wherein i=1,2 ..., N, (N+1) are the sum of second point；

In embodiments of the present invention, the stretching module includes:

Second acquisition unit, for obtaining the location information of j-th of pixel in i-th of subregion；

Third determination unit, for according to the location information of j-th of pixel, central point, first point of concentration I-th first point, i+1 first point, the second point i-th of second point, the i+1 second point, i-th concentrated Adjustment point and i+1 adjustment point determine stretching function；

4th determination unit, for being determined according to the location information of j-th of pixel and the stretching function J-th of target position；

5th determination unit, for determining the object pixel of j-th of pixel according to j-th of target position Value；

Updating unit, for the pixel value of j-th of pixel to be updated to the target pixel value, after obtaining processing Image.

In embodiments of the present invention, the third determination unit, comprising:

Extend subelement, for arriving the 4th line of the central point and j-th of pixel along the central point The direction of j-th of pixel is extended, with described i-th first point, the i+1 first point of line intersects at First intersection point mutually gives the second intersection point with the line of i-th of second point, the i+1 second point, connects with the described 5th Line intersects at third intersection point, wherein the 5th line is the line of the i-th adjustment point with i+1 adjustment point；

6th determines subelement, for determining stretching function according to the 4th distance, the 5th distance and the 6th distance, In, the 4th distance is the distance between the central point and first intersection point, and the 5th distance is the central point The distance between described second intersection point, the 6th distance is the distance between the central point and the third intersection point.

In embodiments of the present invention, the described 6th determine that subelement is also used to:

The first coordinate is determined according to first ratio and second ratio；

In embodiments of the present invention, the 4th determination unit, comprising:

7th determine subelement, for according to the location information of j-th of pixel determine j-th of pixel with The 7th distance between the central point；

8th determines subelement, determines the third ratio between the 7th distance and the 6th distance；

Computation subunit, it is described for being calculated using the third ratio as the input of the stretching function The corresponding output valve of third ratio；

9th determines subelement, for determining the 8th distance according to the output valve and the 6th distance, wherein described 8th distance is the distance between j-th of target position and described central point；

Tenth determines subelement, for the position according to the 8th distance, the central point and j-th of pixel Information determines j-th of target position.

In embodiments of the present invention, the 5th determination unit, comprising:

11st determines subelement, is integer for the coordinate value in response to the target position, by the target position Pixel value be determined as the target pixel value of j-th of pixel；

12nd determines subelement, for not being integer in response to the coordinate value of the target position, according to preset algorithm Determine the corresponding pixel value in the target position；The corresponding pixel value in the target position is determined as j-th of pixel Target pixel value.

The embodiment of the present invention provides a kind of image processing equipment, and described image processing equipment includes at least: memory, communication Bus and processor, in which:

The memory, for storing image processing program；

The communication bus, for realizing the connection communication between processor and memory；

The processor, for obtaining the image processing program stored in memory by the communication bus, to execute The step of image processing method as described above.

The embodiment of the present invention provides a kind of storage medium, and image processing program, the figure are stored on the storage medium The step of image processing method as described above is realized when being executed by processor as processing routine.

The embodiment of the present invention provides a kind of image processing method and its device, equipment and storage medium, wherein obtains first Q key point in facial image, wherein Q is the integer greater than 1；Then filtered out from the Q key point default to The key point of processing, this makes it possible to the calculation amounts of less image procossing；And then further according to the default key to be processed Point determines target area to be processed in facial image, includes the forehead region of facial image in the target area；And by institute It states target area and is divided into N number of subregion, N is the integer more than or equal to 2；Finally respectively to the pixel in each subregion Point carries out stretching, the image that obtains that treated；In this way, leading to too small amount of key point to be processed determines target area, And target area is divided into multiple subregions, then stretching is individually carried out to each subregion, it can not only be fitted various The three-dimensional forehead of various kinds achievees the effect that three-dimensional U.S. face, so that treated image is more naturally, and also simplify realization side Formula is to greatly improve treatment effeciency.

Detailed description of the invention

Figure 1A is the application scenarios schematic diagram of image processing method of the embodiment of the present invention；

Figure 1B is the schematic diagram of the another application scenarios of the embodiment of the present invention；

Fig. 2 is the implementation process schematic diagram of image processing method of the embodiment of the present invention；

Fig. 3 A is the implementation process schematic diagram of image processing method of the embodiment of the present invention；

Fig. 3 B is the schematic diagram that the embodiment of the present invention carries out tri patch when forehead plentiful adjustment；

Fig. 3 C is the schematic diagram that the embodiment of the present invention carries out tri patch when forehead reduces adjustment；

Fig. 3 D is the coordinate schematic diagram that the embodiment of the present invention carries out piecewise function when forehead plentiful adjustment；

Fig. 3 E is the coordinate schematic diagram that the embodiment of the present invention carries out piecewise function when forehead reduces adjustment；

Fig. 4 is the implementation process schematic diagram of image processing method of the embodiment of the present invention；

Fig. 5 is the schematic diagram of the fitting broken line of the original forehead of the embodiment of the present invention and target forehead profile；

Fig. 6 is the schematic diagram that the embodiment of the present invention is split as forehead region multiple continuous tri patch；

Fig. 7 is the comparison diagram that the embodiment of the present invention carries out tri patch stretching when forehead full adjustment；

Fig. 8 is the adjustment effect comparison diagram that the embodiment of the present invention carries out facial image when forehead full adjustment；

Fig. 9 A is the comparison diagram that the embodiment of the present invention carries out tri patch stretching when the adjustment of forehead area reduction；

Fig. 9 B is the adjustment effect comparison diagram that the embodiment of the present invention carries out facial image when the adjustment of forehead area reduction；

Figure 10 is the composed structure schematic diagram of image processing apparatus of the embodiment of the present invention；

Figure 11 is the composed structure schematic diagram of image processing equipment of the embodiment of the present invention.

Specific embodiment

In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention In attached drawing, the specific technical solution of invention is described in further detail.The following examples are intended to illustrate the invention, but does not have to To limit the scope of the invention.

Figure 1A is the application scenarios schematic diagram of image processing method of the embodiment of the present invention, as shown in Figure 1A, the application scenarios In include terminal 101 and server 102, terminal 101 and server 102 establish communication connection by network and go forward side by side row information friendship Mutually.Terminal 101 obtains image 1011 to be processed first, includes facial image region in image 1011, then will be to be processed Image 1011 is sent to server 102, and the image processing method provided according to embodiments of the present invention by server 102 is to be processed Image 1011 handled, the image 1012 that obtains that treated, then will treated that image 1012 is sent to terminal, by terminal 101 output display treated images 1012.Wherein, image to be processed can be adopted in real time by the image collecting device of terminal Collect image, can also be the image for being stored in terminal local.

Figure 1B is the schematic diagram of the another application scenarios of the embodiment of the present invention, only includes in application scenarios as shown in Figure 1B Terminal 111, terminal 111 obtain image 1111 to be processed first, include facial image region in image 1111, then terminal 111 image processing methods provided according to embodiments of the present invention handle image 1111 to be processed, obtain that treated Image 1112, finally again by terminal output display treated image 1112.

The application scenarios schematic diagram in conjunction with shown in Figure 1A and Figure 1B, below to image processing method and image processing apparatus, Each embodiment of equipment is illustrated.

The present embodiment provides a kind of image processing method, Fig. 2 is the implementation process of image processing method of the embodiment of the present invention Schematic diagram, as shown in Fig. 2, this method comprises:

Step S201 obtains Q key point in facial image.

Here, step S201 can be by terminal realization, can also be by server realization.In the present embodiment, Q For the integer greater than 1.Facial image can be the image including forehead region shot under different angle, for example can be people The facial image of face face camera lens, can be with lateral face to the facial image of camera lens.

When step S201 is realized by terminal, facial image can be to be adopted in real time using the image collecting device in terminal The image collected, such as can be preview image when taking pictures using camera applications, it can also be the face of terminal local storage Image, facial image can be the image shot by camera applications at this time, can also be the face downloaded from network Image, naturally it is also possible to be the facial image that other terminals are sent to the terminal, when the face figure that user stores terminal local When as being unsatisfied with, it can use image processing method provided in an embodiment of the present invention and being adjusted.

When step S201 is realized by server, facial image can be terminal and be sent to server, at this time in order to Guarantee the good experience of taking pictures of user, facial image is generally the image being locally stored in terminal.

Step S201 can be the human face region identified in facial image by face recognition technology when realizing Characteristic point namely Q key point.In the present embodiment, the characteristic point that forehead region is included at least in Q key point, can be with Characteristic point including regions such as eye, nose, mouth, chins.

Step S202 filters out default key point to be processed from Q key point.

Here, step S202 can be by terminal realization, can also be by server realization.

In the present embodiment, the characteristic point including forehead region in a key point to be processed, therefore, step S202 are preset When realizing, the characteristic point that forehead region is filtered out from Q key point can be.In this way, due to being screened from multiple key points A small amount of key point to be processed out, to reduce the calculation amount of image procossing, and then can be improved image processing efficiency.

Step S203 determines target area to be processed in the facial image according to the default key point to be processed Domain.

Here, step S203 can be by terminal realization, can also be by server realization.In target area at least Forehead region including facial image.

Due to including the characteristic point in forehead region in the default key point to be processed, step S203 when realizing, Can with when first the characteristic point in forehead region is carried out carrying out interpolation obtaining more key points (the first point set), and according to adjustment Ratio-dependent adjusts distance, then is adjusted more key points that interpolation obtains are carried out according to adjustment distance, is adjusted Multiple key points (the second point set) afterwards, and multiple key points adjusted are sequentially connected, determine target area.

Target area is divided into N number of subregion by step S204.

Here, step S204 can be by terminal realization, can also be by server realization.

In a kind of enforceable mode, subregion can be triangular shape.It include multiple passes on the profile of target area Key point, when target area is divided into N number of subregion, can be will be on the central point of target area and the profile of target area Each key point (adjustment point) be sequentially connected to obtain N number of subregion.

In the present embodiment, N is the integer greater than 2；Embedded the first sub- triangular facet with same apex angle of each subregion Piece and the second sub- tri patch, wherein the bottom edge of the first sub- tri patch is two neighboring pass on the original contour in forehead region The line of key point (first point), the bottom edge of the second sub- tri patch are two neighboring key point on the objective contour in forehead region The line of (second point), two ends on the bottom edge of two endpoints and the second sub- tri patch on the bottom edge of the first sub- tri patch Point is on tri patch by the point on two bevel edges on vertex.

Step S205 carries out stretching to the pixel in each subregion respectively, the image that obtains that treated.

Here, step S205 can be by terminal realization, can also be by server realization.

Pixel progress stretching in subregion may be considered, the pixel value of pixel is replaced with into another picture The pixel value of vegetarian refreshments, wherein another pixel mentioned here can be the central point along target area to the side of the pixel The pixel that certain distance obtains is stretched out to by pixel, can also be pixel along the direction of the pixel to central point The point pixel that retraction certain distance obtains inwards.It can be realized by carrying out stretching to the pixel in subregion to volume The stretching of head region profile or retraction, to realize the adjustment in forehead region.

Step S205 can be and be clicked through using preset stretching algorithm to the pixel in each subregion when realizing Row stretching, in the present embodiment, preset stretching algorithm are linear function, therefore have linear computation complexity, Efficiency faster, is adapted to camera live preview function to the U.S. efficient demand of face algorithm.

When step S205 is realized by terminal, after step S205, this method further include: after exporting and showing processing Image；When step S205 is realized by server, after step S205, this method further include: after server will be handled Image be sent to terminal；Terminal shows and exports the image that receives that treated.

In image processing method provided in an embodiment of the present invention, Q key point in facial image is obtained first, then Default key point to be processed is filtered out from the Q key point, this makes it possible to the calculation amounts for reducing image procossing；Into And target area to be processed in facial image is determined further according to the default key point to be processed, in the target area Forehead region including facial image；And the target area is divided into N number of subregion, N is the integer more than or equal to 2；Most Stretching is carried out to the pixel in each subregion respectively afterwards, the image that obtains that treated；In this way, leading to too small amount of Key point to be processed determines target area, and target area is divided into multiple subregions, can not only be fitted various each The three-dimensional forehead of sample achievees the effect that three-dimensional U.S. face, so that treated, image is more natural；In addition, to each subregion It individually carries out stretching and uses the stretching function with linear computation complexity, implementation can be simplified, and And computational efficiency is high, to greatly improve treatment effeciency.

Based on the above embodiments, the embodiment of the present invention provides a kind of image processing method again, is applied to terminal, and Fig. 3 A is The implementation process schematic diagram of image processing method of the embodiment of the present invention, as shown in Figure 3A, this method comprises:

Step S301, terminal obtain Q key point in facial image.

Here, Q is the integer greater than 1.Facial image can be to be collected in real time using the image collecting device in terminal Image, such as can be preview image when taking pictures using camera applications, can also be the facial image of terminal local storage, Facial image can be the image shot by camera applications at this time, can also be the facial image downloaded from network, It can certainly be the facial image that other terminals are sent to the terminal, when the facial image that user stores terminal local is discontented When meaning, it can use image processing method provided in an embodiment of the present invention and be adjusted.

Step S301 can be the human face region identified in facial image by face recognition technology when realizing Characteristic point namely Q key point.In the present embodiment, the characteristic point that forehead region is included at least in Q key point, can be with Characteristic point including regions such as eye, nose, mouth, chins.

Step S302, terminal filter out default key point to be processed from Q key point.

Here, in the present embodiment, it includes each on the original contour of forehead region for presetting in a key point to be processed Characteristic point, therefore, step S202 can be from the original contour for filtering out forehead region in Q key point when realizing Characteristic point.In this way, due to filtering out a small amount of key point to be processed from multiple key points, to reduce the meter of image procossing Calculation amount, and then can be improved image processing efficiency.

Step S303, terminal obtain the facial angle information and adjusting parameter of facial image to be processed.

Here, step S303 can be facial image inputting a preset Face datection mould during realization Type, to obtain the facial angle information in facial image, wherein when face is face camera lens, facial angle information is 0 degree, When face is biased to 30 degree of the left side, facial angle information is 30 degree, when face is biased to 30 degree of the right, facial angle information for- 30 degree.

Adjusting parameter can be the real number between 0 to 1, can also be the real number between -1 to 0, wherein adjusting parameter Absolute value is bigger, then adjusts dynamics or adjustment amount is bigger.

When adjusting parameter be positive number when, be to extend to the outside forehead region, adjustment it is fuller, when adjusting parameter is negative It is by forehead region to contract when number.

Step S304, terminal determine in forehead region according to fisrt feature point set, facial angle information and adjusting parameter The second point set on the first point set and objective contour on heart point, original contour.

Here, step S304 can be realized by following steps:

Step S3041 determines the central point of target area according to fisrt feature point set and facial angle information.

Here, step S3041 realize when be utilized fisrt feature point set location information and facial angle information it is true Set the goal the central point in region.

Step S3042 determines second feature point set according to central point, fisrt feature point set and adjusting parameter.

Here, step S3042 can be during realization and be sequentially connected central point and each fisrt feature point, And determine the distance between central point and each fisrt feature point, then an adjustment ratio is determined further according to adjusting parameter Example, and then determine adjustment distance, then the line of central point and fisrt feature point is adjusted according to the adjustment distance, it determines Second feature point.Second feature point set may be considered the set of the characteristic point on the objective contour in forehead region.

Step S3043 carries out interpolation according to preset interpolation algorithm respectively to fisrt feature point set and second feature point set, Accordingly obtain the first point set on original contour and the second point set on objective contour.

Here, due to image processing method provided in an embodiment of the present invention be for forehead plentiful degree and/or What the width at left and right sides of forehead was adjusted, therefore institute's fisrt feature point set and second feature point set are being inserted according to preset When value-based algorithm carries out interpolation, the characteristic point and second feature point set that the fisrt feature point at left and right sides of forehead region is concentrated can be In characteristic point carry out interpolation respectively.Interpolation can be carried out using Catmull-Rom polygon approach method when realizing.

Step S305, terminal is according in central point, the second point set and preset second adjustment ratio-dependent facial image Target area.

Here, the implementation method of step S305 is similar with step S3042, and central point and each second point are sequentially connected, And determine the distance between central point and each second point, then further according to second adjustment ratio-dependent second adjustment distance, The line of central point and second point is adjusted according to the second adjustment distance finally, adjustment point is determined, by each adjustment Point is sequentially connected, and determines target area.

It in embodiments of the present invention, include the original contour and objective contour in forehead region in target area.It needs It is bright, when the second adjustment ratio used when the second adjustment ratio used when reducing forehead region is than plentiful forehead region Greatly, to guarantee the original contour and objective contour in the target area determined including forehead region.

Target area is divided into N number of subregion by step S306, terminal.

Here, N is integer greater than 2, and subregion is the tri patch in other embodiments, therefore this step can also be with It is interpreted as target area being divided into N number of continuous tri patch, embedded the with same apex angle of each tri patch One sub- tri patch and the second sub- tri patch, wherein the bottom edge of the first sub- tri patch is on the original contour in forehead region Two neighboring first point of line, the bottom edge of the second sub- tri patch are two neighboring second point on the objective contour in forehead region Line.

Step S306 can be when realizing is sequentially connected composition i-th for central point, the i-th adjustment point and i+1 adjustment point Sub-regions are to obtain N number of subregion, wherein i=1,2 ..., N+1, (N+1) are the sum of adjustment point.

Step S307, terminal obtain the location information of j-th of pixel in i-th of subregion.

Here, j=1,2 ..., M, wherein M is the pixel sum in i-th of subregion.The position of j-th of pixel Information can be the coordinate of j-th of pixel.

Step S308, terminal is according to the location information of j-th of pixel, central point, i-th the of first point of concentration A bit, i-th of second point, i+1 second point, the i-th adjustment point and the i+1 that first point of i+1, the second point are concentrated Adjustment point determines stretching function.

Here, i-th first point and i+1s first point be the first sub- tri patch bottom edge endpoint, i-th the 2 points and i+1 second point are the endpoint on the bottom edge of the second sub- tri patch.

One stretching function will be determined according to the position of the pixel for each pixel, pass through the flexible change The pixel value of pixel on original contour can be mapped as the pixel value of the pixel on objective contour by exchange the letters number, and should Stretching function is linear function, has linear complexity, therefore treatment effeciency is high.

Step S309, terminal determine j-th of target position according to the location information and stretching function of j-th pixel It sets.

Here, step S309 is during realization, can first according to the location information of j-th of pixel determine one it is defeated Enter value, which is input in stretching function, the corresponding output valve of the input value is obtained, then according to the output valve Determine corresponding j-th of the target position of j-th of pixel.

Step S310, terminal determine the target pixel value of j-th of pixel according to j-th of target position.

Here, when the coordinate value of j-th of target position is integer, the corresponding pixel value in j-th of target position is determined For target pixel value；When the coordinate value of j-th of target position is floating number, j-th of target can be determined according to preset algorithm The corresponding pixel value in position can use bilinear interpolation method and determine the corresponding picture in j-th of target position during realization Element value.

The pixel value of j-th of pixel is updated to target pixel value by step S311, terminal, the image that obtains that treated.

In image processing method provided in an embodiment of the present invention, firstly, terminal obtains Q key in facial image Then point filters out default key point to be processed, in this way according only to a small amount of pass filtered out from the Q key point Key point is handled, it will be able to reduce the calculation amount of image procossing；Then it based on default key point to be processed and obtains again The facial angle information and adjusting parameter got determine the central point in forehead region, the first point set on original contour and target wheel The second point set on exterior feature, and then according to the mesh in central point, the second point set and preset second adjustment ratio-dependent facial image Region is marked, then target area is divided into N number of continuous subregion, for example, target area is divided into N number of tri patch, and The location information of each pixel in each tri patch is obtained, then the location information further according to each pixel, center Point determines a stretching function with, the second point of second point concentration at first point of first point of concentration, and then terminal is further according to three The location information of each pixel and stretching function determine the target pixel value of each pixel in edged surface piece, finally by The pixel value of j pixel is updated to target pixel value, the image that obtains that treated, in this way, due to that can pass through tri patch It is fitted arbitrary polygon, therefore by implementing image processing method provided in this embodiment, various solids can be fitted Forehead achievees the effect that three-dimensional U.S. face, so that treated image is more naturally, and the stretching function in the present embodiment For linear function, there is linear complexity, computation complexity, to greatly improve image processing efficiency.

In other embodiments, step S3042 can be realized by following steps:

Step 21, the first distance between central point and fisrt feature point is determined.

Step 22, the first adjustment distance is determined according to first distance and adjusting parameter.

Here, step 22 realize when, can be and the first adjustment ratio is determined according to adjusting parameter first, then further according to First distance and the first adjustment ratio-dependent the first adjustment distance.

Since adjusting parameter can be positive value, or negative value, thus determine that the first adjustment distance also correspondingly For positive value or negative value.

Step 23, will be determined as accordingly along the first adjustment direction apart from the point that fisrt feature point is the first adjustment distance Second feature point.

Here, the direction for arriving fisrt feature point is put centered on the first adjustment direction.When the first adjustment distance is positive value, step Rapid 23, when realizing, are that fisrt feature point is extended the adjustment distance along the first adjustment direction, obtain second feature point；Assuming that the One apart from being 10, and the first adjustment distance is 2, then being exactly that fisrt feature point will be extended distance 2 and obtained along the first adjustment direction To second feature point, that is to say, that second feature point is by distance center point is 12 along the first adjustment direction point.

When the first adjustment distance is negative value, step 23 is by fisrt feature point along the first adjustment direction phase when realizing The absolute value of anti-direction retraction the first adjustment distance, obtains second feature point；Assuming that first distance is 10, the first adjustment distance It is -2, then being exactly that the point along the contrary direction of the first adjustment at a distance from fisrt feature point for 2 is determined as second Characteristic point, that is to say, that second feature point is by distance center point is 8 along the first adjustment direction point.

Step 24, it obtains the fisrt feature and concentrates the corresponding second feature point of each fisrt feature point, obtain the Two feature point sets.

In other embodiments, step S3043 can be realized by following steps:

Step 241, the first subset of the fisrt feature point set and the third subset of corresponding second feature point set are drawn It is divided into first object subregion, by the 4th subset of the second subset of the fisrt feature point set and corresponding second feature point set It is divided into the second target subregion.

Here, the first subset in first object subregion including fisrt feature point set and the third of second feature point set Collect, includes the second subset of fisrt feature point set and the 4th subset of second feature point set in the second target subregion.

It should be noted that first object subregion and the second target subregion are the region of left and right two in forehead region.

Step 242, interpolation is carried out according to preset interpolation algorithm respectively to the first subset and second subset, obtained corresponding 5th subset and the 6th subset.

Step 243, the intersection of the 5th subset and the 6th subset is determined as the first point set.

Step 244, interpolation is carried out according to preset interpolation algorithm respectively to third subset and the 4th subset, obtained corresponding 7th subset and the 8th subset.

Step 245, the intersection of the 7th subset and the 8th subset is determined as second subset.

In the embodiment where step 241 to step 245, being will be in the sub- target area in left and right two in target area Feature point set carry out interpolation arithmetic respectively, that is to say, that in embodiments of the present invention, there is no to forehead region top edge Profile carries out interpolation arithmetic, i.e., is not changed to the hair line of user, so that more aesthetically pleasing to effect of the forehead after flexible.

In other embodiments, step S305 can be realized by following steps:

Step 31, the second distance between central point and i-th of second point of second point concentration is determined.

Here, i=1,2 ..., N+1, wherein (N+1) is the sum of second point, due to the sum of second point and first point Sum and adjustment point sum it is all identical, so first point sum also be N+1, adjust a little sum also be N+1.

Step 32, according to second distance and second adjustment ratio-dependent second adjustment distance.

Here, second adjustment ratio is positive value.Second adjustment ratio is typically less than 1 real number, such as second adjustment ratio Example can be 0.15.

Step 33, by along second adjustment direction i-th of second point of distance be second adjustment distance point be determined as the i-th tune Integral point.

Here, to the direction of i-th of second point centered on second adjustment direction o'clock；I-th first point in central point and i-th On the second line for adjusting point.

Step 34, target area is determined according to (N+1) a adjustment point.

Here it is possible to which adjustment point adjacent in (N+1) a adjustment point is sequentially connected, target area is determined.

In other embodiments, step S308 can be realized by following steps:

Step 41, the 4th line of central point and j-th of pixel is carried out along central point to the direction of j-th of pixel Extend, with i-th first point, i+1 first point of line intersect at the first intersection point, with i-th of second point, the i+1 The line of a second point intersects at the second intersection point, intersects at third intersection point with the 5th line.

Here, the 5th line is the line of the i-th adjustment point and i+1 adjustment point namely the bottom edge of tri patch.Fig. 3 B is The embodiment of the present invention carries out the schematic diagram of tri patch when forehead plentiful adjustment, and as shown in Figure 3B, A, B, C are respectively tri patch Three vertex, A, D, E be the first sub- tri patch three vertex, A, F, G be the second sub- tri patch three vertex, three Edged surface piece ABC, the first sub- tri patch ADE, the second sub- tri patch AFG share an apex angle.BC is the 5th line, DE the The bottom edge of one sub- tri patch, and DE is a part of forehead region original contour, FG is the bottom edge of the second sub- tri patch, and FG is a part of forehead regional aim profile.Assuming that j-th of pixel is point P, the first intersection point is point I at this time, and second hands over Point is point J, and third intersection point is point K.

Step 42, stretching function is determined according to the 4th distance, the 5th distance and the 6th distance.

Here, point is the distance between with first intersection point centered on the 4th distance, i.e. the distance between AI, during the 5th distance is The distance between heart point and the second intersection point, i.e. the distance between AJ, the 6th puts the distance between third intersection point centered on, That is the distance between AK.

Fig. 3 C is the schematic diagram that the embodiment of the present invention carries out tri patch when forehead reduces adjustment, as shown in Figure 3 C, A, B, C Respectively three vertex of tri patch, A, D, E are three vertex of the first sub- tri patch, and A, F, G are the second sub- triangular facet Three vertex of piece, tri patch ABC, the first sub- tri patch ADE, the second sub- tri patch AFG share an apex angle.BC is 5th line, DE is the bottom edge of the first sub- tri patch, and DE is a part of forehead region original contour, and FG is the second son three The bottom edge of edged surface piece, and FG is a part of forehead regional aim profile.Assuming that j-th of pixel is point P, first hand at this time Point is point I, and the second intersection point is point J, and third intersection point is point K.

In other embodiments, step 42 can be realized by following steps:

Step 421, it determines between the first ratio, the 5th distance and the 6th distance between the 4th distance and the 6th distance Second ratio.

Step 422, the first coordinate is determined according to the first ratio and the second ratio.

Here, the abscissa of the first coordinate is the second ratio, and the ordinate of the first coordinate is the first ratio.

Step 423, the linear equation of the first coordinate and the line of origin is determined as the first piecewise function.

Here, the first piecewise function can use formula (1-1) expression:

Wherein, in formula (1-1), DIS_AI is the 4th distance, and DIS_AJ is the 5th distance.

Step 424, the linear equation of the first coordinate and the line of preset second coordinate is determined as the second piecewise function.

Here, preset second coordinate can be (1,1), and the second piecewise function can be indicated with formula (1-2):

In formula (1-2), DIS_AK is the 6th distance.

Step 425, stretching function is determined according to the first piecewise function and the second piecewise function.

Here, stretching function can be indicated with formula (1-3):

Wherein, in formula (1-3), DIS_AK is the 6th distance.

In the full adjustment of progress forehead, since second point is on central point and first point of extended line, that is to say, that the The area of two sub- tri patch is greater than the first sub- tri patch, and the 5th distance is greater than the 4th distance, at this time the coordinate of piecewise function Schematic diagram is as shown in Figure 3 C, wherein 301 be point represented by the first coordinate, and 302 be point represented by the second coordinate.

When carrying out forehead diminution adjustment, since second point is between central point and first point, that is to say, that the second son three The area of edged surface piece is less than the first sub- tri patch, and the 5th distance is less than the 4th distance, the coordinate schematic diagram of piecewise function at this time As shown in FIGURE 3 E, wherein 303 be point represented by the first coordinate, and 302 be point represented by the second coordinate.

In other embodiments, step S309 can be realized by following steps:

Step 51, according to the location information of j-th of pixel determine between j-th of pixel and central point the 7th away from From.

Here, suppose that j-th of pixel, between central point A and the first intersection I, the 7th distance is indicated with DIS_AP.

Step 52, the third ratio between the 7th distance and the 6th distance is determined.

Here, third ratio is

Step 53, using third ratio as the input of stretching function, output valve is obtained.

Here, output valve

Step 54, the 8th distance is determined according to output valve and the 6th distance.

Here, the 8th distance is the distance between j-th of target position and central point, it is assumed that j-th of target position is P ', 8th distance

Step 55, j-th of target position is determined according to the location information of the 8th distance, central point and j-th of pixel.

Here, it is known that the coordinate of the length of AP ', central point A, and the coordinate of point P being located on AP ' can be determined The coordinate of j-th of target position P '.

In other embodiments, step S310 can be realized by following steps:

Step 61, the coordinate value in response to target position is integer, and the pixel value of target position is determined as j-th of pixel The target pixel value of point；

Step 62, the coordinate value in response to target position is not integer, determines that target position is corresponding according to preset algorithm Pixel value；The corresponding pixel value in target position is determined as to the target pixel value of j-th of pixel.

Here, step 62 can determine the corresponding pixel value in target position when realizing according to bilinear interpolation method, In other embodiments, the pixel of four pixel nearest with target position can also be determined according to the coordinate value of target position Value, then mean value computation is carried out to four pixel values, using obtained mean value as the corresponding pixel value in j-th of target position.

In other embodiments, step 42 can also be realized by following steps:

Step 426, third coordinate is determined according to the 4th distance and the 5th distance；

Step 427,4-coordinate is determined according to the 6th distance；

Step 428, the linear equation of third coordinate and the line of origin is determined as third piecewise function；

Step 429, the linear equation of third coordinate and the line of 4-coordinate is determined as the 4th piecewise function.

Step 430, stretching function is determined according to third piecewise function and the 4th piecewise function.

Here, in the embodiment where step 426 to step 430, the table of third piecewise function and the first piecewise function Be up to formula it is identical, the expression formula of the 4th piecewise function and the second piecewise function be it is identical, what is determined in step 430 is flexible Transforming function transformation function is also stretching function identical, that only step 430 obtains with the stretching function determined in step 425 In stretching function obtained in step 425, the value range of x is different, the stretching function that step 430 obtains It can be indicated with formula (1-4):

Correspondingly, when the method provided using step 426 to step 430 determines stretching function, step S309 can To be realized by following steps:

Step 71, according to the location information of j-th of pixel determine between j-th of pixel and central point the 7th away from From.

Step 72, the input by the 7th distance as stretching function, is calculated output valve.

Here, output valve is the distance between j-th of target position and central point, it is assumed that and j-th of target position is P ', 8th distance

Step 73, j-th of target position is determined according to the location information of the 8th distance, central point and j-th of pixel.

Based on the above embodiments, the embodiment of the present invention provides a kind of image processing method again, and Fig. 4 is the embodiment of the present invention The implementation process schematic diagram of image processing method, as shown in figure 4, method includes the following steps:

Step S401, input face picture and " forehead is full " dynamics parameter.

Here, " forehead is full " dynamics parameter reflects the plentiful degree of forehead after image procossing, " forehead is full " dynamics ginseng Number is bigger, and finally obtained forehead is more plentiful.

Step S402 carries out Face datection to face picture, exports forehead key point and POSE face information.

Here, Face datection is carried out to face picture using preset Face datection model, forehead region can be obtained Key point and POSE facial angle information.Wherein, when face face camera lens, facial angle information is 0 degree, to the left to be positive Angle, the angle to the right being negative.

Step S403 is fitted forehead profile using Catmull-Rom polygon approach algorithm and the forehead key point of input.

Step S404 determines " forehead " objective contour by incoming " forehead is full " dynamics parameter and forehead contour line, Using the objective contour of the POSE attribute amendment all angles of Face datection, the full center of forehead is determined using forehead key point.

Here, the center-filled direction of " forehead is full " is determined using POSE angle information and forehead key point, and then really One filling center of fixed " forehead is full "；Then " forehead is full " filling center and forehead key point are sequentially connected and is extended, The point on extended line is taken according to a certain percentage according to " forehead is full " dynamics parameter to determine " forehead is full " forehead key point afterwards Position；Finally using Catmull-Rom polygon approach method and forehead it is full " afterwards forehead key point carry out interpolation obtain more Point on more forehead profiles connects the fitting broken line of these point composition volume target forehead profiles.

Forehead region is split as multiple continuous tri patch by step S405.

It here, will be corresponding on each point and target forehead profile on " forehead is full " filling center, original forehead profile The extension certain proportion of line of each point obtain vertex of a triangle set, center and vertex are filled according to " forehead is full " Forehead region is split as multiple continuous tri patch by set, is fitted three-dimensional forehead using continuous tri patch.

Here the subregion in tri patch, that is, other embodiments.

Step S406, for each tri patch, using the stretching formula and bilinear interpolation ground certainly to triangle Pixel in dough sheet does stretching.

Step S407 judges whether to have handled all tri patch.

Here, if having handled all tri patch, S408 is entered step；If not handled all triangular facets Piece enters step S406.

Step S408, the effect picture after output " forehead is full " processing.

Image processing method provided in an embodiment of the present invention is on the basis of a small amount of key point of Face datection model calibration It is upper further to use polygon approach forehead profile using Catmull-Rom polygon approach method.Organic U.S. face camera is to inspection The precision and execution efficiency for surveying model have high requirement, and detection model can be effectively relieved using fitted polygon process Performance pressures.Meanwhile the deformation algorithm used in stretching has linear complexity, efficiency faster, is adapted to camera Live preview function is to the U.S. efficient demand of face algorithm, and the deformation method has certain fault-tolerance, can be directed to forehead A certain range region carries out whole deformation around profile, can slow down crucial point tolerance bring negative effect, overall effect is more Stablize.

Include following three techniqueflows in image processing method provided in an embodiment of the present invention, is to utilize folding respectively Line segment is fitted the profile of forehead and target forehead, forehead region is split as multiple continuous tri patch and based on triangular facet The figure stretching of piece.Wherein:

The first step utilizes the profile of broken line fitting forehead and target forehead.

Here, in this step, forehead key point, the posture (POSE) of face picture, the output of Face datection algorithm are inputted Facial angle information and " forehead is full " dynamics parameter.Export the fitting broken line of forehead and target forehead profile.Fig. 5 is this hair The schematic diagram for being fitted broken line of bright embodiment original forehead and target forehead profile, as shown in figure 5,501 be original forehead profile Fitting broken line, 502 be target forehead profile fitting broken line.

During realization, which can be realized by following steps:

Step 11, the center-filled direction of " forehead is full " is determined using POSE angle information and forehead key point, in turn Determine a filling center of " forehead is full "；

Step 12, " forehead is full " filling center and forehead key point are sequentially connected and is extended, according to " forehead is full " power Degree parameter takes the point on extended line according to a certain percentage to determine the position of " forehead is full " forehead key point afterwards；

Step 13, using Catmull-Rom polygon approach method and the more forehead wheels of new and old forehead key point interpolation Point on exterior feature connects the fitting broken line that these points constitute original forehead and target forehead profile.

In other embodiments, after step 13, it is also necessary to be sequentially connected " forehead is full " filling center and target volume The point obtained after interpolation in head contour and extension take the point on extended line according to a certain percentage to determine adjustment point, successively connect Connect the fitting broken line that adjacent adjustment point constitutes a transition profile.

Forehead region is split as multiple continuous tri patch by second step.

Here, the fitting for being fitted broken line and transition profile of original forehead, target forehead profile is inputted in this step The filling center of broken line and " forehead is full " exports several continuous tri patch.

Fig. 6 is the schematic diagram that the embodiment of the present invention is split as forehead region multiple continuous tri patch, such as Fig. 6 institute Show, be sequentially connected filling center and adjust the fitting broken line of profile.

Third step, the figure stretching based on tri patch.

Here, tri patch and Control point are inputted in this step, the tri patch after exporting stretching.

Fig. 3 B is the signal that the embodiment of the present invention carries out the tri patch and Control point that input when the full adjustment of forehead Figure, as shown in Figure 3B, the stretching for controlling tri patch need altogether seven points, and wherein tri- points of ABC are tri patch Three vertex, four additional point D, E and F, G are the point on original forehead profile and the corresponding broken line of target forehead profile respectively.

The core of stretching is that arbitrary point P in input tri patch is constructed mapping relations P '=f (P), so that defeated The point P of tri patch corresponding position takes the pixel value at the position point P ' out, to complete the shift transformation that point P ' arrives point P.This hair The figure stretching method that bright embodiment provides can deforming tri patch with fast and flexible.The stretching method The picture at other based on control point can also be applied to deform field.

For each pixel P, tie point P in tri patch and extend and DE, FG and BC intersect at I respectively, J and K.Length DIS_AP, DIS_AI, DIS_AJ and the DIS_AK of AP, AI, AJ and AK are found out respectively.Structural segmentation function y=g (x), So that the corresponding piecewise function image passing point (0,0) of y=g (x), (1,1) and (DIS_AI/DIS_AK, DIS_AJ/DIS_ AK).X=DIS_AP/DIS_AK substitution equation y=g (x) is obtained into y=DIS_AP '/DIS_AK (piecewise function refers to Fig. 3 D). In view of point A is filling center, so point P ' should be on line segment AK, and then the point P ' on available line segment AK, so that The length of AP ' is DIS_AP '.

Finally, for arbitrary point P in tri patch, the pixel of the corresponding position point P ' is all taken, to complete stretching. The case where being floating number for point P ' coordinate, bilinear interpolation algorithm can be used and get corresponding pixel value.Fig. 7 is the present invention Embodiment carries out the comparison diagram of tri patch stretching when forehead full adjustment, as shown in fig. 7,701 be original tri patch, 702 be the tri patch after stretching, wherein hatched example areas is to be filled part, and white space is elongated portion.

Fig. 8 is the adjustment effect comparison diagram that the embodiment of the present invention carries out facial image when forehead full adjustment, wherein 801 are Facial image to be processed, 802 be facial image adjusted, by comparing 801 and 802 it is found that people adjusted The forehead of face image is fuller.

Tri patch fitting and stretching method are comprehensively utilized in the image processing method, and to complete camera, " forehead is full It is full " function, to achieve the effect that three-dimensional U.S. face.Three-dimensional full face meet the aesthetic of Asians, and 3D transform can remold volume Head, effect are more natural.

In the diminution adjustment for carrying out forehead region, the tri patch and Control point of input are as shown in Figure 3 C, control The stretching of tri patch needs altogether seven points, and wherein tri- points of ABC are three vertex of tri patch, four additional point D, E and F, G are the point on original forehead profile and the corresponding broken line of target forehead profile respectively, at this time D, E to A distance be greater than F, the distance of G to A.

When carrying out the treatment process of stretching in the diminution adjustment for carrying out forehead region and carry out the full adjustment of forehead Process it is similar, the piecewise function of use is as shown in FIGURE 3 E.When Fig. 9 A is that the embodiment of the present invention carries out the adjustment of forehead area reduction The comparison diagram of tri patch stretching, as shown in Figure 9 A, 901 be original tri patch, and 902 be the triangular facet after stretching Piece, wherein hatched example areas is to be filled part, and blank region is elongated portion.

Fig. 9 B is the adjustment effect comparison diagram that the embodiment of the present invention carries out facial image when the adjustment of forehead area reduction, wherein 911 be facial image to be processed, and 912 be facial image adjusted, after comparing 911 and 912 it is found that adjusting The forehead region of facial image be obviously reduced.

3D " forehead is full " U.S. face can be carried out to human face photo using image processing method provided in an embodiment of the present invention, Firstly, being fitted forehead wheel using Face datection model calibration forehead key point, and using Catmull-Rom polygon approach method It is wide.Later, divide three-dimensional forehead using continuous tri patch according to clockwise direction.Finally, being directed to each triangular facet Piece carries out stretching using stretching algorithm, reaches 3D " forehead is full " effect.In image provided in an embodiment of the present invention In processing method, be fitted complicated, miscellaneous solid forehead using tri patch, then to each tri patch individually into Row deformation algorithm simplifies implementation and greatly improves efficiency of algorithm.

Tri patch fitting, which can be played, breaks the whole up into parts, and simplifies deformation process and quickly establishes the work of 3D mathematical model With can flexibly cope with different angle, the forehead of size and shape.Image processing method tool provided in an embodiment of the present invention simultaneously There is whole the characteristics of deforming in proportion, so that deformation effect is more natural.

The embodiment of the present invention provides a kind of image processing apparatus, and Figure 10 is the composition of image processing apparatus of the embodiment of the present invention Structural schematic diagram, as shown in Figure 10, the image processing apparatus 1000 include: the first acquisition module 1001, screening module 1002, One determining module 1003, division module 1004 and stretching module 1005, in which:

Described first obtains module 1001, for obtaining Q key point in facial image, wherein Q is whole greater than 1 Number；

The screening module 1002, for filtering out default key point to be processed from the Q key point；

First determining module 1003, for according to the default key point to be processed determine in the facial image to The target area of processing, wherein include the forehead region of facial image in target area；

Division module 1004, for target area to be divided into N number of subregion, wherein N is the integer greater than 2；

Stretching module 1005 is obtained for carrying out stretching to the pixel in each subregion respectively Treated image.

In other embodiments, the default key point to be processed includes the of forehead region in the facial image One feature point set, accordingly, the first determining module 1003 includes:

First acquisition unit, for obtaining the facial angle information and adjusting parameter of facial image；

First determination unit, for determining forehead region according to fisrt feature point set, facial angle information and adjusting parameter Central point, the second point set on the first point set and objective contour on original contour；

Second determination unit, for according to central point, the second point set and preset second adjustment ratio-dependent target area.

In other embodiments, the first determination unit, comprising:

First determines subelement, for determining the center of target area according to fisrt feature point set and facial angle information Point；

Second determines subelement, for determining second feature point set according to central point, fisrt feature point set and adjusting parameter；

Interpolation subelement, for being carried out respectively according to preset interpolation algorithm to fisrt feature point set and second feature point set Interpolation accordingly obtains the first point set and the second point set.

In other embodiments, second determine that subelement is also used to:

Determine the first distance between central point and fisrt feature point；

The first adjustment distance is determined according to first distance and adjusting parameter；

It is special by being determined as corresponding second apart from the point that fisrt feature point is the first adjustment distance along the first adjustment direction Levy point, wherein point arrives the direction of fisrt feature point centered on the first adjustment direction；

In other embodiments, interpolation subelement is also used to:

It is first object by the first subset of fisrt feature point set and the third subset division of corresponding second feature point set The second subset of fisrt feature point set and the 4th subset division of corresponding second feature point set are the second target by subregion Region；Interpolation is carried out according to preset interpolation algorithm respectively to the first subset and second subset of first object subregion, is obtained Corresponding 5th subset and the 6th subset；

Third subset and the 4th subset to the second target subregion carry out interpolation according to preset interpolation algorithm respectively, obtain To corresponding 7th subset and the 8th subset；

In other embodiments, the second determination unit, comprising:

Third determines subelement, for determining second between central point and i-th of second point of second point concentration Distance, wherein i=1,2 ..., N+1, (N+1) are the sum of second point；

4th determines subelement, for according to second distance and second adjustment ratio-dependent second adjustment distance；

Adjust subelement, for by along second adjustment direction i-th of second point of distance be second adjustment distance point it is true It is set to the i-th adjustment point, wherein to the direction of i-th of second point centered on second adjustment direction o'clock；

5th determines subelement, for determining target area according to (N+1) a adjustment point.

In other embodiments, division module 1004, comprising:

Division unit, for central point, the i-th adjustment point and i+1 adjustment point to be sequentially connected and constitute i-th of subregion, Wherein, i=1,2 ..., N, (N+1) are the sum of second point；

In other embodiments, stretching module 1005 includes:

Third determination unit, for according to the location information of j-th of pixel, central point, it is described first point concentration i-th A first point, i-th of second point concentrating of i+1 first point, the second point, i+1 second point, the i-th adjustment put and I+1 adjustment point determines stretching function；

4th determination unit, for determining j-th of target according to the location information and stretching function of j-th pixel Position；

5th determination unit, for determining the target pixel value of j-th of pixel according to j-th of target position；

Updating unit, for the pixel value of j-th of pixel to be updated to target pixel value, the image that obtains that treated.

In other embodiments, third determination unit, comprising:

Extend subelement, for by the 4th line of central point and j-th of pixel along central point to j-th pixel Direction is extended, with described i-th first point, the i+1 first point of line intersect at the first intersection point, with described the I second point, the i+1 second point line mutually give the second intersection point, with the 5th line intersect at third intersection point, In, the 5th line is the line of the i-th adjustment point with i+1 adjustment point；

6th determines subelement, for determining stretching function according to the 4th distance, the 5th distance and the 6th distance, In, the distance between point and the first intersection point centered on the 4th distance, the 5th puts the distance between second intersection point centered on, The distance between point and third intersection point centered on 6th distance.

In other embodiments, the 6th determine that subelement is also used to:

Determine the second ratio between the first ratio, the 5th distance and the 6th distance between the 4th distance and the 6th distance Value；

The first coordinate is determined according to the first ratio and the second ratio；

Stretching function is determined according to the first piecewise function and the second piecewise function.

In other embodiments, the 4th determination unit, comprising:

7th determines subelement, for according to the location information of j-th of pixel determine j-th of pixel and central point it Between the 7th distance；

Computation subunit, for it is corresponding that third ratio to be calculated using third ratio as the input of stretching function Output valve；

9th determines subelement, for determining the 8th distance according to output valve and the 6th distance, wherein the 8th distance is the The distance between j target position and central point；

Tenth determines subelement, for determining jth according to the location information of the 8th distance, central point and j-th of pixel A target position.

In other embodiments, the 5th determination unit, comprising:

11st determines subelement, is integer for the coordinate value in response to target position, by the pixel value of target position It is determined as the target pixel value of j-th of pixel；

12nd determines subelement, is not integer for the coordinate value in response to target position, is determined according to preset algorithm The corresponding pixel value in target position；The corresponding pixel value in target position is determined as to the target pixel value of j-th of pixel.

It should be noted that the description of apparatus above embodiment, be with the description of above method embodiment it is similar, have The similar beneficial effect with embodiment of the method.For undisclosed technical detail in apparatus of the present invention embodiment, this hair is please referred to The description of bright embodiment of the method and understand.

It should be noted that if realize above-mentioned image processing method in the form of software function module, and as only Vertical product when selling or using, also can store in a computer readable storage medium.Based on this understanding, originally Substantially the part that contributes to existing technology can be in the form of software products in other words for the technical solution of inventive embodiments It embodies, which is stored in a storage medium, including some instructions are used so that a computer Equipment (can be personal computer, server or network equipment etc.) execute each embodiment method of the present invention whole or Part.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read Only Memory), magnetic disk Or the various media that can store program code such as CD.In this way, the embodiment of the present invention is not limited to any specific hardware It is combined with software.

Correspondingly, the embodiment of the present invention provides a kind of computer storage medium again, is stored in the computer storage medium Computer executable instructions, the computer executable instructions realize image procossing provided by the above embodiment when being executed by processor The step of method.

Correspondingly, the embodiment of the present invention provides a kind of image processing equipment, and Figure 11 is that image procossing of the embodiment of the present invention is set Standby composed structure schematic diagram, as shown in figure 11, the image processing equipment 1100 include: a processor 1101, at least one Communication bus 1102, user interface 1103, at least one external communication interface 1104 and memory 1105.Wherein, communication bus 1102 are arranged for carrying out the connection communication between these components.Wherein, user interface 1103 may include display screen, PERCOM peripheral communication Interface 1104 may include the wireline interface and wireless interface of standard.Wherein processor 1101, configuration pass through the communication bus The image processing program that stores in memory is obtained, the step of to execute image processing method provided by the above embodiment.

The description of images above processing equipment and computer storage medium embodiment, the description with above method embodiment are Similar, there is with embodiment of the method similar beneficial effect.For computer program product of the present invention, computer equipment and meter Undisclosed technical detail in calculation machine storage medium embodiment, please refers to the description of embodiment of the present invention method and understands.

It should be understood that " one embodiment " or " embodiment " that specification is mentioned in the whole text mean it is related with embodiment A particular feature, structure, or characteristic is included at least one embodiment of the present invention.Therefore, occur everywhere in the whole instruction " in one embodiment " or " in one embodiment " not necessarily refer to identical embodiment.In addition, these specific features, knot Structure or characteristic can combine in any suitable manner in one or more embodiments.It should be understood that in various implementations of the invention In example, magnitude of the sequence numbers of the above procedures are not meant that the order of the execution order, and the execution sequence of each process should be with its function It can determine that the implementation process of the embodiments of the invention shall not be constituted with any limitation with internal logic.The embodiments of the present invention Serial number is for illustration only, does not represent the advantages or disadvantages of the embodiments.

It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row His property includes, so that the process, method, article or the device that include a series of elements not only include those elements, and And further include other elements that are not explicitly listed, or further include for this process, method, article or device institute it is intrinsic Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do There is also other identical elements in the process, method of element, article or device.

In several embodiments provided herein, it should be understood that disclosed device and method can pass through it Its mode is realized.Apparatus embodiments described above are merely indicative, for example, the division of the unit, only A kind of logical function partition, there may be another division manner in actual implementation, such as: multiple units or components can combine, or It is desirably integrated into another system, or some features can be ignored or not executed.In addition, shown or discussed each composition portion Mutual coupling or direct-coupling or communication connection is divided to can be through some interfaces, the INDIRECT COUPLING of equipment or unit Or communication connection, it can be electrical, mechanical or other forms.

Above-mentioned unit as illustrated by the separation member, which can be or may not be, to be physically separated, aobvious as unit The component shown can be or may not be physical unit；Both it can be located in one place, and may be distributed over multiple network lists In member；Some or all of units can be selected to achieve the purpose of the solution of this embodiment according to the actual needs.

In addition, each functional unit in various embodiments of the present invention can be fully integrated in one processing unit, it can also To be each unit individually as a unit, can also be integrated in one unit with two or more units；It is above-mentioned Integrated unit both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.

Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above method embodiment can pass through The relevant hardware of program instruction is completed, and program above-mentioned can store in computer-readable storage medium, which exists When execution, step including the steps of the foregoing method embodiments is executed；And storage medium above-mentioned includes: movable storage device, read-only deposits The various media that can store program code such as reservoir (Read Only Memory, ROM), magnetic or disk.

If alternatively, the above-mentioned integrated unit of the present invention is realized in the form of software function module and as independent product When selling or using, it also can store in a computer readable storage medium.Based on this understanding, the present invention is implemented Substantially the part that contributes to existing technology can be embodied in the form of software products the technical solution of example in other words, The computer software product is stored in a storage medium, including some instructions are used so that computer equipment (can be with It is personal computer, server or network equipment etc.) execute all or part of each embodiment the method for the present invention. And storage medium above-mentioned includes: various Jie that can store program code such as movable storage device, ROM, magnetic or disk Matter.

The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.

Claims

1. a kind of image processing method, which is characterized in that the described method includes:

Obtain Q key point in facial image, wherein Q is the integer greater than 1；

Default key point to be processed is filtered out from the Q key point；

Target area to be processed in the facial image, the target area are determined according to the default key point to be processed It include the forehead region of facial image in domain；

The target area is divided into N number of subregion, N is the integer more than or equal to 2；

2. image processing method according to claim 1, which is characterized in that the default key point packet to be processed The fisrt feature point set in forehead region in the facial image is included, it is described that people is determined according to the default key point to be processed Target area to be processed in face image, comprising:

The central point, original in the forehead region is determined according to the fisrt feature point set, facial angle information and adjusting parameter The first point set on profile and the second point set on objective contour；

3. image processing method according to claim 2, which is characterized in that it is described according to the fisrt feature point set, Facial angle information and adjusting parameter determine the central point in the forehead region, the first point set and objective contour on original contour On the second point set, comprising:

Interpolation is carried out according to preset interpolation algorithm respectively to the fisrt feature point set and the second feature point set, it is mutually deserved To first point set and the second point set.

4. image processing method according to claim 3, which is characterized in that it is described according to the central point, described One feature point set and adjusting parameter determine second feature point set, comprising:

The point that the point of fisrt feature described in the distance along the first adjustment direction is the first adjustment distance is determined as corresponding the Two characteristic points, wherein point arrives the direction of fisrt feature point centered on the first adjustment direction；

It obtains the fisrt feature and concentrates the corresponding second feature point of each fisrt feature point, obtain second feature point set.

5. image processing method according to claim 3, which is characterized in that described to the fisrt feature point set and institute It states second feature point set and carries out interpolation according to preset interpolation algorithm respectively, accordingly obtain first point set and the second point set, Include:

It is first object by the first subset of the fisrt feature point set and the third subset division of corresponding second feature point set The second subset of the fisrt feature point set and the 4th subset division of corresponding second feature point set are the second mesh by subregion Mark subregion；

To first subset of the first object subregion and the second subset respectively according to preset interpolation algorithm into Row interpolation obtains corresponding 5th subset and the 6th subset；

The third subset and the 4th subset to the second target subregion respectively according to preset interpolation algorithm into Row interpolation obtains corresponding 7th subset and the 8th subset；

6. image processing method according to claim 2, which is characterized in that described according to the central point, second point Target area described in collection and preset second adjustment ratio-dependent, comprising:

Determine the second distance between the central point and i-th of second point of second point concentration, wherein i=1,2 ..., N+1, (N+1) are the sum of second point；

It is that the point of the second adjustment distance is determined as the i-th adjustment by i-th of second point described in the distance along second adjustment direction Point, wherein the second adjustment direction is the central point to the direction of i-th of second point；

The target area is determined according to (N+1) a described adjustment point.

7. image processing method according to claim 6, which is characterized in that described that the target area is divided into N Sub-regions, comprising:

8. a kind of image processing apparatus, which is characterized in that described image processing unit include: the first acquisition module, screening module, First determining module, division module and stretching module, in which:

First determining module, it is to be processed in the facial image for being determined according to the default key point to be processed Target area, wherein in the target area include facial image forehead region；

The division module, for the target area to be divided into N number of subregion, wherein N is the integer more than or equal to 2；

The stretching module is handled for carrying out stretching to the pixel in each subregion respectively Image afterwards.

9. a kind of image processing equipment, which is characterized in that described image processing equipment includes at least: memory, communication bus and Processor, in which:

The memory, for storing image processing program；

The processor, for obtaining the image processing program stored in memory by the communication bus, with perform claim It is required that the step of image processing method described in 1 to 7 any one.

10. a kind of storage medium, which is characterized in that be stored with image processing program, described image processing on the storage medium The step of image processing method described in any one of claim 1 to 7 is realized when program is executed by processor.