WO2017054651A1

WO2017054651A1 - Method and device for determining fusion coefficient

Info

Publication number: WO2017054651A1
Application number: PCT/CN2016/099290
Authority: WO
Inventors: 陈岩; 黄英; 邹建法
Original assignee: 阿里巴巴集团控股有限公司; 陈岩; 黄英; 邹建法
Priority date: 2015-09-29
Filing date: 2016-09-19
Publication date: 2017-04-06
Also published as: CN106558043B; CN106558043A

Abstract

A method and device for determining a fusion coefficient. The method comprises: positioning a feature point for a target in an image (101), wherein the feature point comprises a contour point; dividing the target isometrically by utilizing the contour point to form N isometric lines, wherein N is a pre-set positive integer (102); determining a fusion coefficient of each isometric line of an unknown region in the target respectively (103); and performing interpolation processing on each of the other pixel points of the unknown region by utilizing the fusion coefficient of each isometric line of the unknown region to obtain a fusion coefficient of each pixel point in the unknown region (104). The method can reduce the algorithm complexity and improve the real-time performance.

Description

Method and device for determining fusion coefficient

[Technical Field]

The present invention relates to the field of computer application technologies, and in particular, to a method and apparatus for determining a fusion coefficient.

【Background technique】

With the increasing popularity of smart terminals, the demand for image processing using smart terminals is increasing, and various beauty apps are widely favored by beauty lovers. However, the existing APPs are based on the beauty processing of still images. The main method is to locate the key points of the organs in the static images, determine the areas that need to be beautiful according to the positioning results, and use image fusion technology and color probability. The model performs background blending. In the background fusion, it is necessary to determine the fusion coefficient of the unknown region. The unknown region is the transition region between the foreground region and the background region, which is a linear fusion of the foreground region and the background region, and satisfies the relationship represented by the following expression:

U=F×alpha+(1-alpha)×B

Where U represents the matrix of pixel values of the unknown region, F represents the matrix of pixel values of the foreground region, B represents the matrix of pixel values of the background region, and alpha is the fusion coefficient.

The existing fusion coefficient is determined by combining the pixel values of all unknown regions into a super-large linear equation group, and the solution is required to solve the inverse matrix of a super-large matrix. The algorithm is complex, takes a long time, and has poor real-time performance.

[Summary of the Invention]

In view of this, the present invention provides a method and apparatus for determining a fusion coefficient, so as to reduce algorithm complexity and improve real-time performance.

The specific technical solutions are as follows:

The present invention provides a method of determining a fusion coefficient, the method comprising:

Feature point location for an object in the image, the feature point including a contour point;

Using the contour points to divide the target into equal proportions to form N isometric lines, the N a preset positive integer;

Determining a fusion coefficient for each of the equal proportion lines of the unknown region in the target;

The other coefficients of the unknown regions are interpolated by using the fusion coefficients of the respective isometric lines of the unknown region, and the fusion coefficients of the pixels in the unknown region are obtained.

According to a preferred embodiment of the present invention, the target is equally divided by the contour point to form N isometric lines including:

Proportional splitting on the line connecting the center point of the target to the contour point, and obtaining N dividing points on each connecting line;

A proportional line is formed by equal proportions of the points on each line.

According to a preferred embodiment of the present invention, if the contour point includes only the outer contour point, the center point of the target is the center point of the area covered by the target;

If the contour point includes an inner contour point and an outer contour point, the center point of the target is an intermediate point of the inner contour point and its corresponding outer contour point.

According to a preferred embodiment of the present invention, the method further includes:

A foreground area, a background area, and an unknown area of the target are determined.

According to a preferred embodiment of the present invention, if the target has an inner and outer contour, a region consisting of a point between a length of the inner contour and a distance between the inner and outer contours at a point between the first threshold and the second threshold is used as a foreground. The first threshold is smaller than the second threshold, and the first threshold and the second threshold are preset positive numbers less than one;

In the area covered by the target, an area other than the foreground area is regarded as an unknown area;

An area other than the area covered by the target is used as the background area.

According to a preferred embodiment of the present invention, if the target has only an outer contour, a point is formed from a point at which the ratio of the length of the center point of the target to the distance between the center point and the outer contour is within a third threshold. a region, as a foreground region, the third threshold is a preset positive number less than one;

According to a preferred embodiment of the present invention, each of the equal proportions of the unknown region in the target is divided When determining the fusion coefficient, perform the respective proportional lines for the unknown area:

Pixel value sampling of the foreground area point and the background area point respectively along the normal direction for the points on the proportional line;

Determining a fusion coefficient of the target point by using the pixel value of the point and the pixel value of the sampled foreground area point and the background area point;

The fusion coefficients of the contour lines are determined using the fusion coefficients of the points on the proportional lines.

According to a preferred embodiment of the present invention, determining the fusion coefficients of the proportional lines by using the fusion coefficients of the points on the proportional line includes:

Among the fusion coefficients of the points on the proportional line, the fusion coefficient having the most occurrences is used as the fusion coefficient of the proportional lines.

According to a preferred embodiment of the present invention, determining the fusion coefficients of the targeted points by using the pixel values of the points and the pixel values of the sampled foreground region points and the background region points include:

For each point on the equal-scale line, the sampled foreground area point and the background area point are combined, and each combination includes a foreground area point and a background area point, the number of the combinations being less than a preset positive integer;

The fusion coefficients of the points are calculated separately using each combination.

According to a preferred embodiment of the invention, the fusion coefficients for the points are calculated in the following manner:

The alpha is a fusion coefficient for a point, the U is a pixel value of a point, and the F and B are pixel values of a foreground region point and a pixel value of a background region point, respectively.

According to a preferred embodiment of the invention, the number of combinations is greater than 1 and less than 10.

The present invention also provides an apparatus for determining a fusion coefficient, the apparatus comprising:

a positioning unit, configured to perform feature point positioning on an object in the image, where the feature point includes a contour point;

a dividing unit, configured to form an N proportional line by using the contour point to divide the target into equal proportions, wherein the N is a preset positive integer;

a first determining unit, configured to respectively determine a fusion system for each of the equal proportion lines of the unknown region in the target number;

The second determining unit is configured to perform interpolation processing on the other pixels of the unknown region by using the fusion coefficients of the respective isometric lines of the unknown region to obtain a fusion coefficient of each pixel in the unknown region.

According to a preferred embodiment of the present invention, the splitting unit specifically executes:

A proportional line is formed by equal proportions of the points on each line.

According to a preferred embodiment of the present invention, the apparatus further includes: a region dividing unit configured to determine a foreground region, a background region, and an unknown region of the target.

According to a preferred embodiment of the present invention, the area dividing unit specifically performs:

If the target has an inner and outer contour, an area formed by a point between a length of the inner contour and a distance between the inner and outer contours at a point between the first threshold and the second threshold, as the foreground region, the first threshold is smaller than The second threshold, and the first threshold and the second threshold are preset positive numbers less than one;

If the target has only an outer contour, an area formed by a point that is within a third threshold from a distance between a center point of the target and a distance between the center point and the outer contour is used as a foreground area, The third threshold is a preset positive number less than one;

According to a preferred embodiment of the present invention, the first determining unit specifically executes:

According to a preferred embodiment of the present invention, when the first determining unit determines the fusion coefficient of the proportional line by using the fusion coefficient of each point on the proportional line, the first determining unit performs:

According to a preferred embodiment of the present invention, the first determining unit specifically performs: when determining the fusion coefficient of the point to be:

According to a preferred embodiment of the present invention, the first determining unit calculates the fusion coefficient of the point for the following manner:

It can be seen from the above technical solution that the present invention determines the fusion coefficient of the isometric line by means of equal division of the target, and uses the fusion coefficients of the equal proportion lines of the unknown region to the other pixels of the unknown region. Interpolation processing is performed to obtain a fusion coefficient of each pixel in the unknown region. In other words, first replace the point with a line, and then use the interpolation method to avoid the inverse matrix of the large matrix. Solving, both in scale and in algorithm, reduces complexity, takes less time, and improves real-time performance.

[Description of the Drawings]

FIG. 1 is a flowchart of a main method according to an embodiment of the present invention;

2 is a flowchart of a detailed method according to an embodiment of the present invention;

3a and 3b are respectively a schematic diagram and a schematic diagram of a process of equal division according to an embodiment of the present invention;

4 is a schematic diagram of area division according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of pixel value collection along a normal direction according to an embodiment of the present invention; FIG.

FIG. 6 is a diagram showing the effect of each pixel point fusion coefficient of the lip according to an embodiment of the present invention; FIG.

FIG. 7 is a structural diagram of an apparatus for determining a fusion coefficient according to an embodiment of the present invention.

【detailed description】

The present invention will be described in detail below with reference to the drawings and specific embodiments.

The invention is used for determining a fusion coefficient of an unknown region for an object in an image, the fusion coefficient being used for background fusion of the target, that is, the color model of the target is merged with the background. FIG. 1 is a flowchart of a main method according to an embodiment of the present invention. As shown in FIG. 1 , the method mainly includes the following steps:

In 101, feature point positioning is performed on a target in the image.

The feature points involved in this step mainly include contour points, which may include only outer contour points, for example, the target is an eye; and may include inner contour points and outer contour points, for example, the target is the lips (ie, the inner and outer boundaries of the lips). Depending on the specific target type, some special feature points may also be included. For example, if the target is a lip, the feature point may also include a corner of the mouth. This step actually determines the position information of these feature points, which can be embodied as coordinate information in the image.

In 102, the target is equally divided by the contour points to form N isometric lines, and N is a preset positive integer.

In this step, the profiling is performed on the connecting line from the center point of the target to each contour point, and N dividing points are obtained on each connecting line, and the equal proportioning lines are formed by the equal proportioning points on the respective connecting lines. The main use in the present invention is a proportional line that falls within an unknown area.

Specifically, it can be divided into two cases. If the contour point obtained by the above positioning only includes the outer contour point, the center point is the center point of the area covered by the target, which is better understood. That is, the distance is split from the line connecting the target center point to the outer contour point, for example, divided into 8 equal parts, and there are 7 split points on each line, assuming that the line is cut according to the center point. The order of the starting line number, from 1 to 7 for the number, then all the line points with the number 1 is connected to form a proportional line, all the line points with the number 2 is divided The connections form an isometric line, and so on.

If the contour points obtained by the above positioning include an inner contour point and an outer contour point, the center point is an intermediate point of the inner contour point and its corresponding outer contour point. Usually, the number of inner contour points and outer contour points obtained by the same is the same, and is stored in the form of an array, assuming that there are 10 inner contour points and 10 outer contour points, and the inner contour points of the inner contour point and the outer contour point having the sequence number 1 are The middle point of the inner contour point and the outer contour point of sequence number 2, and so on. The intermediate points are proportionally divided with the corresponding inner contour points, and then the equal proportioning points on the respective connecting lines form a proportional line; the intermediate points are proportionally divided with the corresponding outer contour points, and then The equal proportioning points of the respective lines form an equal proportion line. This situation will be described in detail in the subsequent embodiments.

In 103, the fusion coefficients are determined for each of the isometric lines of the unknown region in the target.

In the embodiment of the present invention, it is considered that the fusion coefficients of each pixel on each isometric line are equal, and other pixels on the unknown region can be interpolated by the fusion coefficients of the proportional lines, thereby avoiding one by one for all points. Determining the fusion coefficient greatly reduces the amount of calculation and improves the calculation efficiency. The manner in which the fusion coefficients of the respective proportional lines are determined will be described in detail in the subsequent embodiments.

In 104, each pixel of the unknown region is interpolated by using the fusion coefficients of the respective isometric lines of the unknown region, and the fusion coefficients of the pixels in the unknown region are obtained.

The manner provided by the present invention can be applied to, for example, color processing an organ in an image, such as coloring lips, eyes, and the like in an image, thereby achieving the purpose of beauty. The following example is in the figure Taking the lips in the image as an example, the fusion coefficient of the unknown region of the lips is determined. FIG. 2 is a flowchart of a detailed method according to an embodiment of the present invention. As shown in FIG. 2, the process may specifically include the following steps:

In 201, feature points are positioned on the lips in the image. The feature points may include: an inner contour and an outer contour of the upper lip, and an inner contour and an outer contour of the lower lip, and may further include a corner of the mouth. For the open or closed state of the lips, the feature point positioning described above can be employed. For the closed state of the lips, it is also possible to position only the outer contour of the upper and lower lips, and may further include positioning of the corners of the mouth.

The method for locating feature points is not limited in the present invention, and any feature point positioning method such as positioning based on SDM (Supervised Descent Method) model and id-exp model positioning may be adopted, and finally the above feature points may be obtained. Location information.

In 202, an equal-split is performed on a line connecting the intermediate point of the inner and outer contour points to the outer contour point, and the equal-distance line is formed by the equal-distinct points on the respective lines, at an intermediate point from the inner and outer contour points. The equidistant division is performed on the line connecting the inner contour points, and the equal proportion lines are formed by the equal-distinct points on the respective connecting lines.

As an example, as shown in Fig. 3a, it is assumed that the positioning results in 12 inner and outer contour points, respectively, the inner contour points are identified as a1 to a12 in Fig. 3a, and the outer contour points are identified as b1 to b12 in Fig. 3a. In the case of equal proportioning, a5 and b5 are taken as examples. A5 and b5 are corresponding inner and outer contour points, and the middle point is o5, and the line connecting o5 and a5 is equally divided, and it is assumed that the section is divided into three sections. There are 2 split points, and the lines connecting o5 and b5 are equally divided. It is assumed that the split is divided into 3 segments and there are 2 split points. The other internal and external contour points are equally divided, taking a6 and b6 as examples, o6 is the middle point of a6 and b6, there are also two split points between o6 and a6, and there are also two cuts between o6 and b6. Points. Connect the split points on the a5 and o5 lines to the split points on the line connecting a6 and o6 one by one. As shown in Figure 3a, these lines are equal proportion lines. Then all the inner and outer contour points are similarly processed, and the resulting equal proportion line can be as shown in Fig. 3b.

In 203, the foreground, background, and unknown regions of the lips in the image are determined.

If the target has an inner and outer contour, assuming the open lips as shown in Figure 3a, the distance can be An area formed by a point between the length of the inner contour and the distance between the inner and outer contours at a point between the first threshold and the second threshold as the foreground area, that is, the intermediate position in the area covered by the target as the foreground region. The first threshold is smaller than the second threshold, and the first threshold and the second threshold are preset positive numbers less than 1, and usually take an empirical value, for example, the first threshold is 0.3, and the second threshold is 0.6. Among the areas covered by the target, areas other than the foreground area are regarded as unknown areas. The area outside the area covered by the target is used as the background area.

If the target has only the outer contour, an area composed of points whose distance from the center point of the target and the distance between the center point and the outer contour is within a third threshold is used as the foreground area, wherein the third threshold is A positive number less than one is usually set, and an empirical value is usually taken. For example, the third threshold is 0.6. Among the areas covered by the target, areas other than the foreground area are regarded as unknown areas. The area outside the area covered by the target is used as the background area.

Taking the lip shown in FIG. 4 as an example, the ratio of the distance between the length of the inner contour and the inner and outer contours on the line 2 is 0.3, and the distance between the point on the line 1 and the length of the inner contour and the inner and outer contours. The ratio is 0.6, the area between line 1 and line 2 is the foreground area, the area between the inner contour and line 2, and the area between the outer contour and line 1 are unknown areas, and the other areas are background areas. For the limitation of subsequent sampling, the range of the background area may be further limited, for example, the range of the background area is limited to: the distance between the length of the outer contour and the inner and outer contours of each point in the background area is smaller than the fourth threshold, fourth The threshold is a preset value between 0 and 1, for example, 0.1. If the distance between the length of the outer contour and the inner and outer contours is 0.1, the area between the line 3 and the outer contour is Background area.

The present invention does not limit the order of the

above steps

202 and 203, and may be performed sequentially in any order, or may be performed simultaneously.

At 204, steps 2041 to 2043 are performed for respective isometric lines of the unknown region, respectively. The following steps 2041 to 2043 are actually the process of determining the fusion coefficients of the isometric lines for each isometric line.

In 2041, pixel values of the foreground area point and the background area point are sampled along the normal direction for the points on the equal-scale line.

In this step, all the pixels on the equal-scale line may be used, or some pixels on the equal-scale line may be selected, for example, some pixels may be selected from the interval on the proportional line.

In 2042, the fusion coefficient of the point is determined using the pixel value of the point and the pixel values of the sampled foreground area and background area points.

For the sampled foreground area points and the background area points can be combined, each combination includes a foreground area point and a background area point, and each combined value is substituted into the following formula, and a fusion coefficient can be calculated:

Where alpha is the fusion coefficient of the point, U is the pixel value of the point, and F and B are the pixel values of the foreground area point and the pixel value of the background area point, respectively.

In order to control the amount of calculation, here, the number of combinations obtained by sampling can be controlled to be between 1 and 10.

As an example, as shown in FIG. 5, the proportional line n is one of the proportional lines in the unknown region, and pn1 is a point on the proportional line at which the background region is collected along the normal direction. Points pb1, pb2, and pb3, foreground region points pf1, pf2, and pf3, after combining the two pairs, obtain the following combinations {pb1, pf1}, {pb1, pf2}, {pb1, pf3}, {pb2, pf1}, { Pb2, pf2}, {pb2, pf3}, {pb3, pf1}, {pb3, pf2}, {pb3, pf3}, and the fusion coefficients of pn1 were obtained by using each combination, and there were nine results. You can take the average of these 9 results, or take the median value, or take one of the values to get the fusion coefficient of the pn1 point. It is also possible to use these 9 results as the fusion coefficients of the pn1 point and continue the subsequent steps.

In 2043, the fusion coefficients of the contour lines are determined using the fusion coefficients of the points on the proportional lines.

In this step, the number of occurrences of each fusion coefficient may be counted in a fusion coefficient of each point on the proportional line, such as a histogram, and the fusion coefficient having the most occurrences is used as the fusion coefficient of the proportional line. Of course, in addition to this method, the fusion coefficient of the contour line may be determined by using the mean value, the median value, and the like of the fusion coefficients of the points on the proportional line.

By performing the above steps 2041 to 2043 for each of the equal proportion lines in the unknown area, you can obtain The fusion coefficient of each proportional line in the unknown region.

In 205, the other coefficients of the unknown region are interpolated by using the fusion coefficients of the respective isometric lines of the unknown region, and the fusion coefficients of the pixels in the unknown region are obtained.

After the fusion coefficient is determined as described above, if the brightness of each pixel of the lip is represented by the brightness degree, the pixel fusion coefficient of the lip portion can be as shown in FIG. 6, and the brighter the color (the lighter the color, the more White) indicates that the fusion coefficient is larger.

After obtaining the fusion coefficient of each pixel in the unknown region, the fusion coefficient can be used to fuse the color model and the current color of the target unknown region in the image to obtain the target after the color processing, and the embodiment is reflected in the lip. If the beauty application colors the lips, the color model to be loaded is blended with the current color of the unknown region of the lips in the image, and the color of the unknown portion of the lips after the beauty is obtained. When color processing, the formula used can be:

Pho=Model×alpha+(1-alpha)×Cur

Where Pho is the pixel value after color processing, Model is the pixel value of the color model to be loaded, and Cur is the current pixel value in the image.

FIG. 7 is a structural diagram of an apparatus for determining a fusion coefficient according to an embodiment of the present invention. The apparatus may be used in an application for color processing an image. As shown in FIG. 7, the apparatus may include: a positioning unit 01, a splitting unit. 02. The first determining unit 03 and the second determining unit 04 may further include a region dividing unit 05. The main functions of each component are as follows:

The positioning unit 01 is responsible for performing feature point positioning on the target in the image, wherein the feature point mainly includes the contour point, and may only include the outer contour point, for example, the target is an eye; and may also include an inner contour point and an outer contour point, for example, the target is a lip ( That is, the inner and outer boundaries of the lips). Depending on the specific target type, some special feature points may also be included. For example, if the target is a lip, the feature point may also include a corner of the mouth.

The splitting unit 02 is responsible for using the contour points to divide the target into equal proportions to form N proportional lines, and N is a preset positive integer. Specifically, the proportional split can be performed on the connecting line from the center point of the target to the contour point, and N dividing points are obtained on each connecting line; the equal-scale lines are formed by the equal-divided points on the respective connecting lines.

Wherein, if the contour point includes only the outer contour point, the center point of the target is the center point of the area covered by the target. That is, the distance is split from the line connecting the target center point to the outer contour point, for example, divided into 8 equal parts, and there are 7 split points on each line, assuming that the line is cut according to the center point. The order of the starting line number, from 1 to 7 for the number, then all the line points with the number 1 is connected to form a proportional line, all the line points with the number 2 is divided The connections form an isometric line, and so on.

If the contour point includes an inner contour point and an outer contour point, the center point of the target is the intermediate point of the inner contour point and its corresponding outer contour point. Usually, the number of inner contour points and outer contour points obtained by the same is the same, and is stored in the form of an array, assuming that there are 10 inner contour points and 10 outer contour points, and the inner contour points of the inner contour point and the outer contour point having the sequence number 1 are The middle point of the inner contour point and the outer contour point of sequence number 2, and so on. The intermediate points are proportionally divided with the corresponding inner contour points, and then the equal proportioning points on the respective connecting lines form a proportional line; the intermediate points are proportionally divided with the corresponding outer contour points, and then The equal proportioning points of the respective lines form an equal proportion line. For a specific example, reference may be made to the description of FIG. 3a and FIG. 3b in the above method embodiment.

The area dividing unit 05 is responsible for determining the foreground area, the background area, and the unknown area of the target. Specifically, if the target has an inner and outer contour, the area dividing unit 05 sets a region composed of a point between the first threshold and the second threshold as a ratio of the distance between the length of the inner contour and the inner and outer contour as the foreground region, A threshold is less than the second threshold, and the first threshold and the second threshold are preset positive numbers less than one, and generally take an empirical value, for example, the first threshold is 0.3, and the second threshold is 0.6. Among the areas covered by the target, areas other than the foreground area are regarded as unknown areas. The area outside the area covered by the target is used as the background area.

If the target has only the outer contour, the region dividing unit 05 sets a region composed of a point whose distance from the center point of the target and the distance between the center point and the outer contour is within a third threshold as the foreground region, and third. The threshold is a preset positive number less than 1, for example, the third threshold is 0.6. Among the areas covered by the target, areas other than the foreground area are regarded as unknown areas. The area outside the area covered by the target is used as the background area.

The first determining unit 03 is responsible for determining the fusion system for each of the isometric lines of the unknown region in the target. number.

Specifically, the first determining unit 03 may first perform pixel value sampling of the foreground area point and the background area point along the normal direction for the points on the equal-scale line. It can be used for all pixels on the equal-scale line, or for some pixels on the equal-scale line. For example, some pixels can be selected from the interval on the proportional line.

Then, the first determining unit 03 determines the fusion coefficient of the point for which the pixel value of the point is used and the pixel value of the sampled foreground area point and the background area point. For the sampled foreground area points and the background area points can be combined, each combination includes a foreground area point and a background area point, and each combined value is substituted into the following formula, and a fusion coefficient can be calculated:

Where alpha is the fusion coefficient of the point, U is the pixel value of the point, and F and B are the pixel values of the foreground area point and the pixel value of the background area point, respectively. In order to control the amount of calculation, here, the number of combinations obtained by sampling can be controlled to be between 1 and 10.

Finally, the first determining unit 03 determines the fusion coefficient of the proportional line by using the fusion coefficients of the points on the equal-scale line. The number of occurrences of each fusion coefficient may be counted in a fusion coefficient of each point on the proportional line, such as a histogram, and the fusion coefficient having the most occurrences is used as a fusion coefficient of the proportional line. Of course, in addition to this method, the fusion coefficient of the contour line may be determined by using the mean value, the median value, and the like of the fusion coefficients of the points on the proportional line.

The second determining unit 04 is responsible for performing interpolation processing on the other pixels of the unknown region by using the fusion coefficients of the respective isometric lines of the unknown region, and obtaining the fusion coefficients of the pixels in the unknown region.

After obtaining the fusion coefficient of each pixel in the unknown region by using the above-mentioned device for determining the fusion coefficient, the color processing application can use the fusion coefficient to fuse the color model with the current color of the target unknown region in the image to obtain the color processing. aims.

In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division, and the actual implementation may have another division manner.

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

The above-described integrated unit implemented in the form of a software functional unit can be stored in a computer readable storage medium. The above software functional unit is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to perform the methods of the various embodiments of the present invention. Part of the steps. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like, which can store program codes. .

The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalents, improvements, etc., which are made within the spirit and principles of the present invention, should be included in the present invention. Within the scope of protection.

Claims

A method for determining a fusion coefficient, the method comprising:

Feature point location for an object in the image, the feature point including a contour point;

And dividing the target into equal proportions by using the contour points to form N isometric lines, wherein the N is a preset positive integer;

Determining a fusion coefficient for each of the equal proportion lines of the unknown region in the target;

The other coefficients of the unknown regions are interpolated by using the fusion coefficients of the respective isometric lines of the unknown region, and the fusion coefficients of the pixels in the unknown region are obtained.
The method according to claim 1, wherein the dividing the target by the contour point to form an N proportional line comprises:

Proportional splitting on the line connecting the center point of the target to the contour point, and obtaining N dividing points on each connecting line;

A proportional line is formed by equal proportions of the points on each line.
The method according to claim 2, wherein if the contour point includes only an outer contour point, a center point of the target is a center point of the area covered by the target;

If the contour point includes an inner contour point and an outer contour point, the center point of the target is an intermediate point of the inner contour point and its corresponding outer contour point.
The method of claim 1 further comprising:

A foreground area, a background area, and an unknown area of the target are determined.
The method according to claim 4, wherein if the target has an inner and outer contour, the ratio of the distance between the length of the inner contour and the inner and outer contour is at a point between the first threshold and the second threshold. And the first threshold is smaller than the second threshold, and the first threshold and the second threshold are preset positive numbers less than 1;

In the area covered by the target, an area other than the foreground area is regarded as an unknown area;

An area other than the area covered by the target is used as the background area.
The method according to claim 4, wherein if the target has only an outer contour, a ratio of a length from a center point of the target to a distance between the center point and the outer contour is at a third threshold The area formed by the points inside, as the foreground area, the third threshold is a preset positive number less than 1;

In the area covered by the target, an area other than the foreground area is regarded as an unknown area;

An area other than the area covered by the target is used as the background area.
The method according to any one of claims 1 to 6, wherein when the fusion coefficients are respectively determined for respective isometric lines of the unknown region in the target, respectively, the respective proportional lines of the unknown region are respectively executed:

Pixel value sampling of the foreground area point and the background area point respectively along the normal direction for the points on the proportional line;

Determining a fusion coefficient of the target point by using the pixel value of the point and the pixel value of the sampled foreground area point and the background area point;

The fusion coefficients of the contour lines are determined using the fusion coefficients of the points on the proportional lines.
The method according to claim 7, wherein the fusion coefficient of each of the proportional lines is determined by using a fusion coefficient of each point on the proportional line:

Among the fusion coefficients of the points on the proportional line, the fusion coefficient having the most occurrences is used as the fusion coefficient of the proportional lines.
The method according to claim 7, wherein the determining the fusion coefficients of the targeted points by using the pixel values of the points for the points and the pixel values of the sampled foreground region points and the background region points comprises:

For each point on the equal-scale line, the sampled foreground area point and the background area point are combined, and each combination includes a foreground area point and a background area point, the number of the combinations being less than a preset positive integer;

The fusion coefficients of the points are calculated separately using each combination.
The method according to claim 9, characterized in that the fusion coefficient for the point is calculated in the following manner:

The alpha is a fusion coefficient for a point, the U is a pixel value of a point, and the F and B are pixel values of a foreground region point and a pixel value of a background region point, respectively.
The method of claim 9 wherein the number of combinations is greater than one and less than ten.
An apparatus for determining a fusion coefficient, the apparatus comprising:

a positioning unit, configured to perform feature point positioning on an object in the image, where the feature point includes a contour point;

a dividing unit, configured to form an N proportional line by using the contour point to divide the target into equal proportions, wherein the N is a preset positive integer;

a first determining unit, configured to respectively determine a fusion coefficient for each of the equal proportion lines of the unknown area in the target;

The second determining unit is configured to perform interpolation processing on the other pixels of the unknown region by using the fusion coefficients of the respective isometric lines of the unknown region to obtain a fusion coefficient of each pixel in the unknown region.
The device according to claim 12, wherein the dividing unit specifically executes:

Proportional splitting on the line connecting the center point of the target to the contour point, and obtaining N dividing points on each connecting line;

A proportional line is formed by equal proportions of the points on each line.
The apparatus according to claim 13, wherein if said contour point includes only outer contour points, a center point of said target is a center point of said target covered area;

If the contour point includes an inner contour point and an outer contour point, the center point of the target is an intermediate point of the inner contour point and its corresponding outer contour point.
The apparatus according to claim 12, further comprising: a region dividing unit configured to determine a foreground region, a background region, and an unknown region of the target.
The device according to claim 15, wherein the area dividing unit specifically performs:

If the target has an inner and outer contour, an area formed by a point between a length of the inner contour and a distance between the inner and outer contours at a point between the first threshold and the second threshold, as the foreground region, the first threshold is smaller than The second threshold, and the first threshold and the second threshold are preset positive numbers less than one;

In the area covered by the target, an area other than the foreground area is regarded as an unknown area;

An area other than the area covered by the target is used as the background area.
The device according to claim 15, wherein the area dividing unit specifically performs:

If the target has only an outer contour, an area formed by a point that is within a third threshold from a distance between a center point of the target and a distance between the center point and the outer contour is used as a foreground area, The third threshold is a preset positive number less than one;

In the area covered by the target, an area other than the foreground area is regarded as an unknown area;

An area other than the area covered by the target is used as the background area.
The device according to any one of claims 12 to 17, wherein the first determining unit specifically executes:

Pixel value sampling of the foreground area point and the background area point respectively along the normal direction for the points on the proportional line;

Determining a fusion coefficient of the target point by using the pixel value of the point and the pixel value of the sampled foreground area point and the background area point;

The fusion coefficients of the contour lines are determined using the fusion coefficients of the points on the proportional lines.
The apparatus according to claim 18, wherein the first determining unit performs the specific use of the fusion coefficient of each point on the proportional line to determine the fusion coefficient of the proportional line:

Among the fusion coefficients of the points on the proportional line, the fusion coefficient having the most occurrences is used as the fusion coefficient of the proportional lines.
The apparatus according to claim 18, wherein the first determining unit specifically performs: when determining a fusion coefficient of the targeted point:

For the points on each isometric line, the sampled foreground area points and background area points are grouped And each combination includes a foreground area point and a background area point, the number of the combinations being less than a preset positive integer;

The fusion coefficients of the points are calculated separately using each combination.
The apparatus according to claim 20, wherein said first determining unit calculates a fusion coefficient for the point in question in the following manner:

The alpha is a fusion coefficient for a point, the U is a pixel value of a point, and the F and B are pixel values of a foreground region point and a pixel value of a background region point, respectively.
The device of claim 20 wherein the number of combinations is greater than one and less than ten.