WO2016107259A1

WO2016107259A1 - Image processing method and device therefor

Info

Publication number: WO2016107259A1
Application number: PCT/CN2015/093026
Authority: WO
Inventors: 李嵩
Original assignee: 努比亚技术有限公司
Priority date: 2014-12-31
Filing date: 2015-10-28
Publication date: 2016-07-07
Also published as: CN104537608A

Abstract

Disclosed are an image processing method and a device therefor. The method comprises: decomposing an input human body photograph according to limb areas to obtain an image of at least one limb area; setting a target posture parameter aiming at each limb area within at least one limb area; according to the target posture parameter of each limb area, selecting a template drawing matching the target posture parameter from template drawings of the limb area; and synthesizing the template drawings matching the target posture parameters in at least one limb area and the images of the limb areas obtained by the decomposition of the human body photograph one by one, and then performing outputting.

Description

Image processing method and device thereof

Technical field

The present invention relates to the field of information technology, and in particular, to a method and an apparatus for image processing.

Background technique

As the saying goes, "Everyone who loves beauty has it." With the advancement of image processing technology, more and more photo beautification software such as beauty and skin are coming into view. However, these softwares mainly focus on the image processing of the face, but not the user groups that have requirements for the body. The existing general-purpose image processing software, such as Photoshop, can be modified to achieve the effect of photo-shaping by the professional operation of some technicians, but requires the technician to have high professionalism, complicated operation, and inconvenience to the user.

At present, there is no method or application software that can automatically beautify the characters in the picture to achieve a slim and robust effect.

Summary of the invention

The invention provides a method and a device for image processing, so as to achieve the purpose of automatically beautifying the figure of a person in a photo, and overcome the defects that the above-mentioned existing image beautification method is complicated and inconvenient for the user to use.

The technical solution of the present invention to solve the above technical problems is as follows.

According to an aspect of the present invention, a method of image processing is provided, the method being applied to an apparatus capable of performing image processing, the apparatus pre-existing at least one template map for at least one limb region of a human body, in the template map At least one set of posture parameters is included, and the method includes:

Decomposing the input human body image according to the limb region to obtain an image of at least one limb region;

Setting a target body state parameter for each of the at least one limb region;

Selecting from the template map of the limb region according to the target body state parameter of each limb region Taking a template map matching the target body state parameter;

A template map matching at least one of the limb regions and the target body state parameter and an image of the limb region obtained by decomposing the human body photograph are combined and output.

According to another aspect of the present invention, an apparatus for image processing is provided, the apparatus comprising:

a photo decomposition module, configured to decompose the input human body photo according to the limb region to obtain an image of at least one limb region;

a parameter receiving module, configured to set a target body state parameter for each of the at least one limb region;

a matching module, configured to select, according to the target body state parameter of each limb region, a template map that matches the target body state parameter from a template map of the limb region;

The synthesizing module is configured to synthesize and output the template image matching the target body state parameter in the at least one limb region and the image of the limb region obtained by decomposing the human body photo.

The invention provides a method and a device for image processing, which automatically synthesize a pre-stored template image and a human body photo input by a user, so that the user can create a desired person's body shape without cumbersome operations, and satisfies The photo of the human body is automatically shaped and beautified. Since the synthetic effect can be adjusted at any time according to the target posture parameters selected by the user, the template map can also include obese and funny template maps, so the user can synthesize healthy and beautiful photos, and can also synthesize obese and quirky photos, which increases the fun. Sex.

DRAWINGS

1 is a flow chart of a method of image processing in accordance with one embodiment of the present invention;

2 is a flow chart of a method of image processing in accordance with another embodiment of the present invention;

3 is a flow chart showing a detailed implementation of a method of image processing in accordance with an embodiment of the present invention;

4 is a block diagram showing an exemplary structure of an apparatus for image processing according to an embodiment of the present invention.

detailed description

The principles and features of the present invention are described in the following with reference to the accompanying drawings.

Embodiment 1

1 is a flow chart of a method of image processing according to an embodiment of the present invention. A method of image processing according to an embodiment of the present invention, which is applied to an apparatus capable of image processing, including but not limited to a mobile phone, a digital camera, a tablet, a computer, etc., is described below with reference to FIG. At least one template map pre-stored in at least one limb region, the template graph including at least one set of posture parameters, the method comprising the steps of:

S100. Decompose the input human body photo according to the limb region to obtain an image of at least one limb region.

The human body photo is divided into an arm region and a trunk region, and the method of decomposing can decompose the limb region in the human body photograph according to different characteristic values corresponding to different limb regions of the human body to obtain an image of at least one limb region.

Wherein, the characteristic value may be a slope of the contour line;

The decomposition may be performed by calculating a hierarchical range of the slope of the outer contour line of the human body region; for example, the same or similar limb regions of the slope may be the same limb region.

S200: Set a target posture parameter for each of the at least one limb region.

The specific implementation manner may be: receiving target posture parameters of different limb regions input by the user.

S300. Select, according to the target body state parameter of each limb region, a template map that matches the target body state parameter from a template map of the limb region.

That is to say, according to the target body state parameter, the template map with the highest degree of matching with the target body state parameter is obtained from the template map of the pre-stored corresponding limb region.

Among them, by querying the target body state parameter input by the user and which group of body parameters are pre-stored in the system Corresponding to the number, and then querying the corresponding template graph group, and obtaining a template graph with the highest matching degree from the template graph group as a synthetic object;

S400. Synthesize and output the template image matching the target body state parameter in at least one limb region and the image of the limb region obtained by decomposing the human body photo.

The template image matching the target body state parameter is selected from the template image of the limb region according to the target body state parameter of each limb region, including:

Obtaining, in each of the target body state parameters of each limb region, pixel values that meet the first preset condition;

Calculating a first total number of pixel values in the target body state parameter of the limb region that meet the first preset condition;

Calculating, according to the first total number, a second total number of pixel values that meet the first preset condition in the body state parameter of the preset template image, and calculating a matching degree between the target body state parameter of the limb region and the template image;

If the matching degree of the target body state parameter of the limb region to the template map is within a preset threshold range, the corresponding template map is selected as a template map matching the target body state parameter of the limb region.

The formula for calculating the above matching degree is:

In this formula, B _s is a value of a pixel belonging to a limb region in a human body photograph (pixels in the limb region are recorded as 1 in the photo, not recorded as 0), and B _t is a value of the pixel in the template map belonging to the limb region ( In the template diagram, the pixel of the limb region is recorded as 1, not recorded as 0), and the sum function represents the sum of the number of 1-value points in the human body photograph, and s is the calculated degree of matching.

Preferably, the above synthesis may include: synthesis of contour lines, and/or fusion of brightness.

Further, the mapping correspondence before and after the synthesis is determined by a grid-based contour method.

The above fusion may be: obtaining the first part of the pixel of the limb region in the photo of the human body The average brightness value is obtained, and the second average brightness value of the pixel in the template image is obtained; and the brightness value of the image after the fusion is calculated based on the preset fusion ratio, and the first average brightness value and the second average brightness value.

The following formula can be used:

V _dst =r*V _t +(1-r)*V _s ;

In this formula, r is the fusion ratio determined according to the user's input, V _t is the brightness value of a pixel in the matched template image, V _s is the brightness value of the human body photo in the corresponding pixel, and V _dst is the fusion Brightness value.

The above target posture parameters include, but are not limited to, at least one of the following: the degree of obesity of the human body after deformation, the degree of muscle display, and the like.

In this embodiment, by using the target body state parameter input by the user to the body state parameter pre-stored in the system, the template image of the corresponding group is filtered, and then a template image with the highest matching degree is obtained in the template image, and the template image is obtained. Synthesize with the human body image input by the user to achieve the desired effect.

Preferably, the embodiment is further capable of re-synthesis. Specifically, the method further includes:

When receiving the rematching instruction, prompting the user to input the target body state parameter again;

After receiving the target body state parameter input by the user again, the above steps are repeated: according to the received target body state parameter, the template image with the highest matching degree is obtained from the template image of the pre-stored corresponding limb region, and the template image with the highest matching degree and the human body input by the user are obtained. The photo is synthesized and output.

The input of the instruction may be displaying a prompt box for the user, prompting the user whether to select re-synthesis or ending the operation through the prompt box; and determining the specified input according to the user's selection.

Since the template gallery can store both robust and slim body types as well as obese and funny body types, you can have more choices when you input the target body parameters. You can synthesize healthy and beautiful photos or synthesize them. Obese, quirky photos add to the fun.

Embodiment 2

2 is a flowchart of a method for image processing according to another embodiment of the present invention. As shown in FIG. 2, the method further includes: Steps S100, S200, S300, and S400.

S210. Identify a skin color region in a photo of a human body.

S220: Perform similar transformation on the graphic of the skin color region, so that the skin color region and the size and position of the limb region in the template image are consistent.

The order of the steps S210 and S220 is not limited to between the steps S300 and S400, and may be between S200 and S300.

Optionally, the method further includes:

When receiving the rematching instruction, prompting the user to input the target posture parameter again;

The similarity transformation includes, but is not limited to, rotating, panning, and scaling the image according to the feature point and the centroid.

The above centroid is the average coordinate point of all pixel coordinates of the connected region, and the coordinate calculation formula is:

Where n represents the number of pixels in the area map.

In this embodiment, the similarity of the photo input by the user is matched with the photo in the template library, and the limb region in the photo is similar in size to the limb region in the template image, so that the limb region in the photo corresponds to the template image. The limb area is better matched.

Embodiment 3

3 is a flow chart showing a detailed implementation of a method of image processing in accordance with an embodiment of the present invention. As shown in Figure 3, its implementation details include the following steps:

S001 performs skin color area detection. A confidence interval for coloring the skin color is obtained by counting the skin color of the human body, and the user input image is segmented by the interval, and the pixels falling within the interval are considered to be the skin color region points.

S002 performs preliminary posture estimation on the skin color region, determines the aspect ratio of the skin color detection region, and determines the direction of the human body in the figure, and if necessary, rotates the image to be the human body standing upright. Use face detection technology to determine whether the photo of the human body is front or back.

S003 uses the geometric distribution of the human body to divide the rough area on the premise that the human body is front and back and upright.

S004 performs feature point detection using edge, texture, and color information in the estimated area. The limb area treated in this embodiment includes an arm, a torso, and both legs. The characteristic points of the specific detection arm area include shoulders, armpits and wrists. The characteristic points of the trunk area include the maximum curvature of the neck and shoulder joints, the nipple and the navel, and the characteristic points of the legs are groin, knee and ankle.

S005 uses the feature detection result to judge the correctness of the attitude estimation. If the feature point detection is correct and the matching degree is high, the posture detection is considered correct; otherwise, the posture detection is performed again. S002~S005 loop repeatedly. If all the gestures cannot detect the feature points correctly, the composition is discarded and the user is prompted to fail.

On the basis of detecting the feature points, S006 divides the human body region into more than the neck, the trunk, the arm, and the leg. Above the neck, the hands and feet are not treated.

S007 performs similar transformation on the divided regions, and transforms the position information of the feature points into a unified manner with the template through the regional centroid.

S008 temporarily stores the transformed area map.

S009 collects the synthetic parameters selected by the user, including but not limited to: gender, fatness, muscle level, shadow depth.

S010 Template Gallery, Template Gallery 041 is a photo gallery, depending on the region (arm, The torso), different genders, different body parameters (fat and thinness, different muscle strength, etc.) are classified into various image templates of different postures of different limbs. The template image may be a grayscale image, the non-human body part gray level is 0, and the human body part gray level is represented as 1. The 042 template screening unit finds a plurality of templates that meet the conditions according to the input limb region, gender, and target body state parameters.

S011 selects a subset of the matching parameters from the template library according to the synthetic parameters selected by the user.

S012 uses the region map transformed by S007 to perform matching on the template images selected by S011 in sequence, and the matching formula is described in the embodiment, and details are not described herein again.

S013 performs an inverse transformation of the matched template on the S007 step similarity transformation to obtain a template map corresponding to the position in the user input map.

S014 uses the image deformation algorithm to calculate the transformation mapping relationship between the limb region map and the template graph. The deformation algorithm can be used based on the grid and the line segment equivalent algorithm.

S015 transforms the user input graph by using the transform mapping relationship, and uses the interpolation method to draw the pixel color.

S016 converts the color of the limb region to the HSV color space, and fuses the Value (luminance) channel and the template grayscale image. The fusion formula is described in the embodiment and will not be described here.

S017 fills the background vacancy part of the transformation using an image restoration algorithm. The final generated image is displayed on the display device.

In this embodiment, the gesture of the human body input by the user is detected before the matching action, and it is initially determined whether the photo can be merged with the photo in the template library, and the subsequent matching success rate is increased, and the unnecessary operation flow is occupied. system resource.

Embodiment 4

4 is a block diagram showing an exemplary structure of an apparatus for image processing according to an embodiment of the present invention, and an apparatus for image processing according to an embodiment of the present invention is described below with reference to FIG. As shown, the device includes:

The photo decomposition module 01 is configured to decompose the input human body photo according to the limb region to obtain an image of at least one limb region;

The parameter receiving module 02 is configured to set a target body state parameter for each of the at least one limb region;

The matching module 03 is configured to select, according to the target body state parameter of each limb region, a template map that matches the target body state parameter from a template map of the limb region;

The synthesizing module 04 is configured to synthesize and output the template image matching the target body state parameter in at least one limb region and the image of the limb region obtained by decomposing the human body photo.

The photo decomposition module is specifically configured to decompose the limb region in the human body photograph according to different feature values corresponding to different limb regions of the human body to obtain an image of at least one limb region.

The matching module is configured to acquire, respectively, a pixel value that meets a first preset condition in a target body state parameter of each limb region; and calculate a first total number of pixel values that meet the first preset condition in the target body state parameter of the limb region. And calculating, according to the first total number, a second total number of pixel values that meet the first preset condition in the body state parameter of the preset template image, and calculating a matching degree between the target body state parameter of the limb region and the template image And if the matching degree of the target body state parameter of the limb region with the template image is within a preset threshold range, selecting a corresponding template image as a template map matching the target body state parameter of the limb region.

The matching module uses the formula:

To calculate the degree of matching, in which B _s is the value of the pixel in the human body photo belonging to the limb region, B _t is the value of the pixel in the template image belonging to the limb region, and the sum function is used to calculate the 1-value point in the human body photo. The sum of the numbers, s is the degree of matching of the calculations.

The synthesis module is used for synthesizing contour lines, and mapping correspondence before and after synthesis Determined by a grid-based contour method. And/or the synthesizing module is configured to perform fusion of brightness.

The synthesizing module is configured to acquire a first average brightness value of a part of the pixel in the body region in the photo of the human body, and obtain a second average brightness value of the pixel in the template image; based on a preset fusion ratio, and the first The average luminance value and the second average luminance value are used to calculate the luminance value of the image after the fusion. The synthesis module uses the formula:

V _dst =r*V _t +(1-r)*V _s

To calculate the brightness of each pixel after fusion, to highlight the contours of the muscles in the photo, so that the combined photo effect is more realistic. In this formula, r is the fusion ratio determined according to the user's input, V _t is the brightness value of a pixel in the matched template image, V _s is the brightness value of the human body photo in the corresponding pixel, and V _dst is the fusion Brightness value.

Optionally, the apparatus for image processing further includes:

a skin color region identification module for identifying a skin color region in a human body photo;

The similarity transformation module is configured to similarly transform the graphics of the skin color region so that the skin color region is consistent with the size and position of the limb region in the template image.

a prompting module, configured to prompt the user to input the target body state parameter again when receiving the rematching instruction;

The repeating module is configured to repeat the above steps after receiving the target body state parameter input by the user again: according to the received target body state parameter, obtaining the template image with the highest matching degree from the template image of the pre-stored corresponding limb region, and matching the template image with the highest matching degree The human body photo input by the user is synthesized and output.

The invention automatically synthesizes the pre-stored template image and the human body photo input by the user, so that the user can create the body shape effect of the desired person without the cumbersome operation, and satisfies the requirement of the human body photo automatic body shaping beautification. Since the synthetic effect can be adjusted at any time according to the target posture parameter selected by the user, the template map can also include a fat and funny template map, so the user can combine Produce healthy and beautiful photos, and also synthesize obese, quirky photos, adding fun.

In the several embodiments provided by the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. The device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner, such as: multiple units or components may be combined, or Can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, or direct coupling, or communication connection of the components shown or discussed may be indirect coupling or communication connection through some interfaces, devices or units, and may be electrical, mechanical or other forms. of.

The units described above as separate components may or may not be physically separated, and the components displayed as the unit may or may not be physical units, that is, may be located in one place or distributed to multiple network units; Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing module, or each unit may be separately used as one unit, or two or more units may be integrated into one unit; the above integration The unit can be implemented in the form of hardware or in the form of hardware plus software functional units.

A person skilled in the art can understand that all or part of the steps of implementing the above method embodiments may be completed by using hardware related to the program instructions. The foregoing program may be stored in a computer readable storage medium, and the program is executed when executed. The foregoing storage device includes the following steps: the foregoing storage medium includes: a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk. A medium that can store program code.

The foregoing is only a specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art is within the technical scope of the present disclosure. Variations or substitutions are readily conceivable and should be covered by the scope of the invention. Therefore, the scope of the invention should be determined by the scope of the appended claims.

Claims

A method of image processing, the method being applied to an apparatus capable of performing image processing, the apparatus being at least one template map pre-stored for at least one limb region of a human body, wherein the template map includes at least one set of body state parameters, Methods include:

Decomposing the input human body image according to the limb region to obtain an image of at least one limb region;

Setting a target body state parameter for each of the at least one limb region;

Selecting, according to the target body state parameter of each limb region, a template map matching the target body state parameter from a template map of the limb region;

A template map matching at least one of the limb regions and the target body state parameter and an image of the limb region obtained by decomposing the human body photograph are combined and output.
The image processing method according to claim 1, wherein the decomposing the input human body photo according to the limb region to obtain an image of at least one limb region comprises:

According to different characteristic values corresponding to different limb regions of the human body, the limb region in the human body photograph is decomposed to obtain an image of at least one limb region.
The image processing method according to claim 1, wherein a template map having the highest degree of matching in each of the at least one limb region, an image of the limb region obtained by decomposing the human body photograph, and the output are combined The method further includes:

Identifying a skin color region in the human body photo;

The pattern of the skin color region is similarly transformed such that the skin color region coincides with the size and position of the limb region in the template image.
The image processing method according to claim 1, wherein the template map matching the target body state parameter is selected from the template image of the limb region according to the target body state parameter of each limb region, including:

Obtaining images of the target posture parameters of each limb region that meet the first preset condition one by one Prime value

Calculating a first total number of pixel values in the target body state parameter of the limb region that meet the first preset condition;

Calculating, according to the first total number, a second total number of pixel values that meet the first preset condition in the body state parameter of the preset template image, and calculating a matching degree between the target body state parameter of the limb region and the template image;

If the matching degree of the target body state parameter of the limb region to the template map is within a preset threshold range, the corresponding template map is selected as a template map matching the target body state parameter of the limb region.
The image processing method according to claim 4, wherein the limb is calculated based on the first total number and a second total number of pixel values of the preset template map that meet the first preset condition The degree to which the target body state parameters of the region match the template map is calculated based on the following formula:

Where B s is the value of the pixel in the human body photo belonging to the limb region, B t is the value of the pixel in the template image belonging to the limb region, and the sum function is the sum of the number of the 1-value points in the human body photo, s is calculated The degree of matching.
The method of image processing according to claim 1, wherein said synthesizing comprises:

The synthesis of contour lines, the mapping correspondence before and after synthesis is determined by the grid-based contour method.
The method of image processing according to claim 1 or 6, wherein said synthesizing comprises fusion of luminance.
The image processing method according to claim 7, wherein the merging of the brightness is: acquiring a first average brightness value of a part of pixels of the limb region in a human body photograph, and acquiring a second average of pixels in the template image Luminance value; based on a preset fusion ratio, and the first flat The average luminance value and the second average luminance value are used to calculate the luminance value of the image after the fusion.
The method of image processing according to claim 8, wherein the calculation formula is:

V dst =r*V t +(1-r)*V s ;

Where r is the fusion ratio determined according to the user's input, V t is the brightness value of a certain pixel in the matched template picture, V s is the brightness value of the corresponding pixel of the human body photo, and V dst is the brightness value after the fusion.
The method of image processing according to claim 1, wherein the method further comprises:

When receiving the rematching instruction, prompting the user to input the target body state parameter again;

After receiving the target body state parameter input by the user again, the above steps are repeated: according to the received target body state parameter, the template image with the highest matching degree is obtained from the template image of the pre-stored corresponding limb region, and the template image with the highest matching degree and the human body input by the user are obtained. The photo is synthesized and output.
An apparatus for image processing, the apparatus comprising:

a photo decomposition module, configured to decompose the input human body photo according to the limb region to obtain an image of at least one limb region;

a parameter receiving module, configured to set a target body state parameter for each of the at least one limb region;

a matching module, configured to select, according to the target body state parameter of each limb region, a template map that matches the target body state parameter from a template map of the limb region;

The synthesizing module is configured to synthesize and output the template image matching the target body state parameter in the at least one limb region and the image of the limb region obtained by decomposing the human body photo.
The device according to claim 11, wherein the photo decomposition module is specifically configured to decompose the limb region in the human body photograph according to different feature values corresponding to different limb regions of the human body to obtain at least one limb region. Image.
The apparatus of claim 11 wherein said apparatus further comprises:

a skin color region identification module, configured to identify a skin color region in the human body photo;

a similarity transformation module, configured to similarly transform the graphic of the skin color region to make the skin The color area coincides with the size and position of the limb area in the template map.
The apparatus according to claim 11, wherein

a matching module, configured to acquire, respectively, a pixel value that meets a first preset condition in a target body state parameter of each limb region; and calculate a first total number of pixel values that meet the first preset condition in the target body state parameter of the limb region; Calculating a matching degree between the target body state parameter of the limb region and the template image, where the first total number and the second total number of pixel values of the preset template map meet the first preset condition; If the matching degree of the target body state parameter of the limb region and the template image is within a preset threshold range, the corresponding template image is selected as a template image matching the target body state parameter of the limb region.
The device according to claim 14, wherein

Matching module for calculating based on the following formula:

Where B s is the value of the pixel in the human body photo belonging to the limb region, B t is the value of the pixel in the template image belonging to the limb region, and the sum function is the sum of the number of the 1-value points in the human body photo, s is calculated The degree of matching.
The apparatus according to claim 11, wherein the synthesis module is configured to perform synthesis of contour lines, and mapping correspondences before and after synthesis are determined by a grid-based contour method.
The apparatus according to claim 11 or 16, wherein said synthesizing module is configured to perform fusion of luminance.
The apparatus according to claim 17, wherein the synthesizing module is configured to acquire a first average brightness value of a part of pixels of the limb region in a photo of a human body, and obtain a second average brightness value of a pixel in the template image; The preset fusion ratio, and the first average luminance value and the second average luminance value, calculate a luminance value of the image after the fusion.
The apparatus according to claim 18, wherein said synthesis module is configured to perform calculation using the following calculation formula:

V dst =r*V t +(1-r)*V s ;

Where r is the fusion ratio determined according to the user's input, V t is the brightness value of a certain pixel in the matched template picture, V s is the brightness value of the corresponding pixel of the human body photo, and V dst is the brightness value after the fusion.
The apparatus of claim 11 wherein said apparatus further comprises:

a prompting module, configured to prompt the user to input the target body state parameter again when receiving the rematching instruction;

The repeating module is configured to repeat the above steps after receiving the target body state parameter input by the user again: according to the received target body state parameter, obtaining the template image with the highest matching degree from the template image of the pre-stored corresponding limb region, and matching the template image with the highest matching degree The human body photo input by the user is synthesized and output.