WO2022016326A1

WO2022016326A1 - Image processing method, electronic device, and computer-readable medium

Info

Publication number: WO2022016326A1
Application number: PCT/CN2020/103061
Authority: WO
Inventors: 丁蕾; 李静; 徐斌
Original assignee: 深圳市大疆创新科技有限公司
Priority date: 2020-07-20
Filing date: 2020-07-20
Publication date: 2022-01-27

Abstract

An image processing method, an electronic device, and a computer-readable medium. An embodiment of the image processing method comprises: performing frequency-division processing on an original image to obtain a high-frequency image, an intermediate-frequency image, and a low-frequency image; performing intensity suppression on pixels in the high-frequency image that meet a preset condition; correcting pixels in the intermediate-frequency image of which the pixels values are outside a preset numerical range; and performing image fusion on the high-frequency image subjected to the intensity suppression, the corrected intermediate-frequency image, and the low-frequency image to generate a target image. Thus, defects such as facial spots and blackheads in an image can be eliminated, and facial texture information and contrast features can be retained, thereby improving the stereoscopic impression of the face in the image.

Description

Image processing method, electronic device and computer readable medium

technical field

The embodiments of the present application relate to the field of computer technologies, and in particular, to an image processing method, an electronic device, and a computer-readable medium.

Background technique

With the popularization of smart terminals, the facial beauty and skin resurfacing function has been applied more and more. Users typically perform beautification of images to smoothen the skin of the faces presented in the images. The microdermabrasion algorithm, as the basic algorithm of the beautification function, can affect the overall image quality level of the beautified image.

In the prior art, some simple filtering operators are usually used to perform low-frequency filtering on the image, so as to achieve the image effect of facial microdermabrasion and beautification. The texture information and contrast features of the face are obtained, resulting in no three-dimensional sense of the face after microdermabrasion.

SUMMARY OF THE INVENTION

The embodiments of the present application propose an image processing method, an electronic device, and a computer-readable medium, so as to solve the problem that the facial texture information and contrast features of the face are lost in the process of beautifying and dermabrasion in the prior art, and the face after dermabrasion has no stereoscopic effect technical problem.

In a first aspect, an embodiment of the present application provides an image processing method, including:

Perform frequency division processing on the original image to obtain a high-frequency image, an intermediate-frequency image and a low-frequency image;

performing intensity suppression on the pixels that meet the preset conditions in the high-frequency image;

Correcting the pixels whose pixel values are outside the preset value range in the intermediate frequency image;

Image fusion is performed on the intensity-suppressed high-frequency image, the modified intermediate-frequency image, and the low-frequency image to generate a target image.

In a second aspect, an embodiment of the present application provides an electronic device, including: a processor and a memory;

the memory for storing program instructions;

The processor executes the program instructions stored in the memory, and when the program instructions are executed, the processor is configured to perform the following steps:

In a third aspect, an embodiment of the present application provides a computer-readable medium on which a computer program is stored, and when the program is executed by a processor, implements the above image processing method.

The embodiments of the present application provide an image processing method, an electronic device, and a computer-readable medium. By suppressing the intensity of pixels satisfying a preset condition in a high-frequency image obtained after frequency division, the high-frequency texture of the image can be preserved while the high-frequency texture of the image can be preserved. Suppresses high frequency imperfections such as blackheads on the human face. By correcting the pixel points whose pixel values are outside the preset value range in the intermediate frequency image obtained after frequency division, it is possible to remove the dark parts such as facial color spots while retaining the contrast information such as the contour of the face. By image fusion of the high-frequency image after intensity suppression, the modified intermediate-frequency image, and the low-frequency image obtained after frequency division, the facial color spots, blackheads and other defects in the original image can be eliminated, and the facial texture information and The contrast feature enhances the three-dimensionality of the face in the image.

Description of drawings

Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

1 is a flowchart of an embodiment of an image processing method according to the present application;

FIG. 2 is a flowchart of another embodiment of an image processing method according to the present application;

Fig. 3 is the flow chart of the process of dividing the frequency of original image into high frequency image and low frequency image;

Fig. 4 is a flow chart of the process of dividing the frequency of a medium and low frequency image into an intermediate frequency image and a low frequency image;

Fig. 5 is the flow chart of the process of filtering medium and low frequency images;

FIG. 6 is a schematic structural diagram of an embodiment of an electronic device according to the present application.

specific embodiment

The present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the related invention, but not to limit the invention. In addition, it should be noted that, for the convenience of description, only the parts related to the related invention are shown in the drawings.

It should be noted that the embodiments in the present application and the features of the embodiments may be combined with each other in the case of no conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

Please refer to FIG. 1 , which shows a flowchart of an image processing method according to the present application. The image processing method can run on various electronic devices, which may include but are not limited to: servers, smart phones, tablet computers, laptop computers, in-vehicle computers, desktop computers, set-top boxes, wearable devices, and the like.

The flow of the image processing method includes the following steps:

Step 101: Perform frequency division processing on the original image to obtain a high-frequency image, an intermediate-frequency image and a low-frequency image.

In this embodiment, the execution body of the image processing method (such as the above-mentioned electronic device) may first acquire the original image. The original image can be captured by the above-mentioned executive body through an image acquisition device (such as a camera) installed in it, or can be pre-stored locally (such as stored in a local album), or obtained by the above-mentioned executive body from the Internet or other devices. . The original image can be an image to be beautified and skinned, and usually can be an image containing a face area.

After acquiring the original image, the above-mentioned execution body may divide the original image according to three frequency bands of high, middle and low frequencies, thereby obtaining a high-frequency image, an intermediate-frequency image and a low-frequency image. The high frequency information of the original image is mainly reflected in the high frequency image, the intermediate frequency information of the original image is mainly reflected in the intermediate frequency image, and the low frequency information of the original image is mainly reflected in the low frequency image.

In general, different frequency information has different roles in the image structure. The main component of the image is the low-frequency information, which forms the basic gray level of the image and has little effect on the image structure. The intermediate frequency information determines the basic structure of the image and forms the main edge structure of the image. The high-frequency information forms the edges and details of the image, which is a further enhancement of the image content on the intermediate-frequency information. Taking a face image as an example, places with sharp changes in brightness or grayscale, such as blackheads, face texture, etc., are usually reflected in high-frequency images. Face contours, color spots, etc., are usually reflected in intermediate frequency images. The gently changing part of the image, usually manifested in low-frequency images.

In this embodiment, during frequency division, the original image may be divided into a high frequency image and a medium and low frequency image first, and then the intermediate frequency image and the low frequency image may be divided by frequency division of the medium and low frequency image. Alternatively, the original image can also be divided into low-frequency images and medium-high-frequency images first, and then the medium-high-frequency images can be divided into high-frequency images and low-frequency images. The order of frequency division is not limited in this embodiment.

In this embodiment, the above-mentioned execution body may perform frequency division by using various commonly used filtering operators. For example, first-order differential operators can be used, such as Roberts (Roberts) operator, Prewitt (Pruitt) operator, Sobel (Sobel) operator, etc.; second-order differential operators can also be used, such as Laplace (Laplace (Laplace) operator Plath) operator, Kirsh operator, etc. The above method can effectively distinguish high-frequency, intermediate-frequency and low-frequency components in an image, and the filter operator used is not specifically limited in this embodiment.

Step 102: Perform intensity suppression on the pixels in the high-frequency image that satisfy the preset condition.

In this embodiment, the above-mentioned execution subject can perform intensity suppression on the pixels in the high-frequency image that meet the preset conditions, so as to suppress high-frequency defects such as blackheads on the human face while retaining the high-frequency texture. Here, the preset conditions are used to filter out pixels that need to be suppressed in intensity, such as pixels where blackheads and other defects are located. High-frequency defects such as blackheads are often significantly higher than the pixel values of surrounding pixels, so pixels that need loudness suppression can be screened based on the pixel values of the neighborhood.

In some optional implementations of this embodiment, each pixel in the high-frequency image can be used as a target point in turn, and the intensity of the pixels that meet the preset conditions in the high-frequency image is suppressed by the following steps:

The first step is to determine the neighborhood of the target point.

Here, the neighborhood of the target point is the area formed by the pixels in the designated area around the target point. As an example, an area formed by 8 pixels that are adjacent to the target point can be taken as the area of the target point. These 8 pixel points can be respectively the pixel point located at the upper left of the target point, the pixel point located directly above the target point, the pixel point located at the upper right of the target point, the pixel point located on the left side of the target point, and the pixel point located on the right side of the target point. Pixel point, the pixel point at the lower left of the target point, the pixel point directly below the target point, the pixel point at the lower right of the target point. It should be noted that other settings may also be performed on the target point area, which is not limited to the above example.

The second step is to obtain the maximum pixel value and/or the minimum pixel value in the neighborhood.

In the third step, in response to the difference between the pixel value of the target point and the maximum pixel value being greater than the first preset value, or the difference between the minimum pixel value and the pixel value of the target point being greater than the second preset value, the pixel value of the target point Make corrections.

Specifically, if the difference between the pixel value of the target point and the maximum pixel value is greater than the first preset value, it means that the pixel value of the target point is significantly larger than the pixel values of other pixels in its neighborhood, and the target point is a high-frequency defect. The pixel value of the target point can be reduced.

Similarly, if the difference between the minimum pixel value and the pixel value of the target point is greater than the second preset value, it means that the pixel value of the target point is significantly smaller than the pixel values of other pixels in its neighborhood, and the target point is a high-frequency defect. The pixel value of the target point can be increased.

In some optional implementations of this embodiment, in response to the difference between the pixel value of the target point and the maximum pixel value being greater than the first preset value, or the difference between the minimum pixel value and the pixel value of the target point being greater than the second preset value Set the value, the median filter algorithm can be used to filter the pixel value of the target point to obtain the corrected pixel value of the target point. The median filter method is a nonlinear smoothing technique, which sets the gray value of each pixel to the median of all pixel gray values in a certain neighborhood window of the point, so as to achieve the effect of suppressing high-frequency defects. .

Optionally, image enhancement processing or no image enhancement processing may be performed on the high-frequency images.

Step 103 , correcting the pixel points in the intermediate frequency image whose pixel values are outside the preset value range.

In this embodiment, since the intermediate frequency information determines the basic structure of the image and forms the main edge structure of the image, important features such as face contours are usually reflected in the intermediate frequency image. However, dark spots such as stains and acne marks are usually present in the IF image, so some pixels in the IF image need to be corrected.

In this embodiment, for the intermediate frequency image, the numerical range of the pixel value can be preset. For each pixel in the intermediate frequency image, it can be detected whether the pixel value of the pixel is within a preset value range. If it is not within the preset value range, the pixel value can be corrected to be within the preset value range.

In some optional implementations of this embodiment, the foregoing preset numerical value range may have a minimum value. For example, the above-mentioned preset value range is [luma2, +∞]. At this time, for each pixel in the intermediate frequency image, in response to the pixel value of the pixel being less than the minimum value in the preset value range (ie luma2), the pixel value of the pixel can be adjusted to the minimum value (ie luma2) ). In this way, points with low brightness (such as stains, acne marks, etc.) can be removed, and the highlight area of the face can be retained.

In some optional implementations of this embodiment, the foregoing preset numerical range may have a maximum value. For example, the above preset value range is [-∞, luma1]. At this time, for each pixel in the intermediate frequency image, in response to the pixel value of the pixel being greater than the maximum value in the preset value range (ie luma1), the pixel value of the pixel is adjusted to the maximum value (ie luma1) . In this way, the points in other areas with obvious high brightness can be eliminated to make the face smoother.

In some optional implementations of this embodiment, the above-mentioned preset value range may have a maximum value and a minimum value at the same time. For example, the above-mentioned preset value range is [luma2, luma1]. At this time, for each pixel in the intermediate frequency image, in response to the pixel value of the pixel being less than the minimum value in the preset value range (ie luma2), the pixel value of the pixel can be adjusted to the minimum value (ie luma2) ). In response to the pixel value of the pixel point being greater than the maximum value in the preset value range (ie luma1 ), the pixel value of the pixel point is adjusted to the maximum value (ie luma1 ). In this way, the points that make the brightness lower and the points that are obviously too bright can be eliminated, and the highlight area of the face can be retained, making the face smoother.

Step 104: Perform image fusion on the intensity-suppressed high-frequency image, the corrected intermediate-frequency image, and the low-frequency image to generate a target image.

In this embodiment, the execution subject may perform image fusion of the intensity-suppressed high-frequency image, the corrected intermediate-frequency image, and the low-frequency image to generate the target image. The fusion method here can be summation of pixel values by pixel.

For example, the size of the intensity-suppressed high-frequency image, the modified intermediate-frequency image, and the low-frequency image is 64×64, which can be regarded as three 64×64 pixel matrices. The above-mentioned execution body may sum the three pixel matrices, and obtain a new pixel matrix after the summation, and the image corresponding to the new pixel matrix is the target image.

In some optional implementation manners of this embodiment, after obtaining the target image, the above-mentioned execution subject may continue to perform the following steps:

The first step is to perform skin color detection on the original image to determine the skin color confidence of the pixels in the original image.

Among them, the confidence of the skin color of the pixel point can be used to represent the probability that the pixel point is the skin color. The foregoing executive body may determine the confidence level of the skin color of the pixels in the original image in various ways.

As an example, the pixel range of the skin color may be preset. If the pixel value of a certain pixel is within the pixel range, the confidence level of the skin color of the pixel can be set to a certain preset value (such as 1 and 0.9, etc.). If the pixel value of a certain pixel is outside the range of the pixel, the confidence of the skin color of the pixel may be set to another preset value (such as 0 or 0.1, etc.).

As another example, multiple pixel ranges may be preset, and different pixel ranges correspond to different confidence levels of skin color. The above executive body may match the pixel value of each pixel in the original image with each pixel range, and detect the pixel range where each pixel value is located. For each pixel, if the pixel value of the pixel is located in a predetermined pixel range, the skin color confidence corresponding to the pixel range is used as the skin color confidence of the pixel.

It should be noted that, when performing skin color detection on the original image, skin color detection may be performed on the entire original image, or skin color detection may be performed on a local area (such as a face area) in the original image, which is not specifically limited here. When skin color detection is performed on a local area (such as a face area) in the original image, the skin color confidence of each pixel in the background area can be set to a small value (such as 0).

It should be pointed out that the manner of determining the confidence level of the skin color of the pixels in the original image is not limited to the above enumeration, and other manners may also be used, which will not be repeated here.

In the second step, based on the confidence of the skin color of the pixels in the original image, the original image and the target image are image-fused to generate the final image.

It is understandable that the pixels with high confidence in the skin color are usually located in the skin area and need to be retouched. Pixels with low confidence in skin color are usually not located in the skin area, and the original features of the area need to be preserved, and the beauty and skin resurfacing of the area should be avoided as much as possible to ensure the clarity of the non-skin area. Therefore, for a pixel with a high confidence in skin color, a larger weight can be set for the pixel value of the pixel in the target image, so as to achieve the effect of beautifying and smoothing the skin. For pixels with low confidence in skin color, a large weight can be set for the pixel value of the pixel in the original image, and the non-skin area remains clear.

Optionally, for each pixel, the following steps can be performed: first, obtain the pixel value of the pixel from the target image (which can be recorded as the first pixel value), and obtain the pixel of the pixel from the original image. value (can be recorded as the second pixel value). Then, the skin color confidence of the pixel is used as the weight of the first pixel value, and the difference between the preset value (eg, 1) and the above weight is used as the weight of the second pixel value. After that, the first pixel value and the second pixel value are weighted and summed to obtain the final pixel value of the pixel. Thus, the pixel value of each pixel point is obtained. Finally, a final image is generated based on the final pixel value of each pixel.

By detecting the skin color confidence of the pixels in the original image, and further fusing the original image with the target image based on the skin color confidence, the background clarity can be maintained in the case of beautifying and resurfacing the face area.

The method provided by the above-mentioned embodiments of the present application can suppress the blackheads and other heights on the face while retaining the high-frequency texture of the image by suppressing the intensity of the pixels that meet the preset conditions in the high-frequency image obtained after frequency division. frequent flaws. By correcting the pixel points whose pixel values are outside the preset value range in the intermediate frequency image obtained after frequency division, it is possible to remove the dark parts such as facial color spots while retaining the contrast information such as the contour of the face. By image fusion of the high-frequency image after intensity suppression, the modified intermediate-frequency image, and the low-frequency image obtained after frequency division, the facial color spots, blackheads and other defects in the original image can be eliminated, and the facial texture information and The contrast feature enhances the three-dimensionality of the face in the image.

Further reference is made to FIG. 2 , which shows the flow of the image processing method according to the present application. The image processing method includes the following steps:

Step 201: Perform frequency division processing on the original image to obtain a high frequency image and a medium and low frequency image.

In this embodiment, the execution body of the image processing method may first perform filtering processing on the original image through a filtering operator to obtain a high-frequency image. Then, based on the high frequency image and the original image, the medium and low frequency image is obtained. Here, various common filtering operators can be used to filter the original image. For example, the Roberts operator, the Prewitt operator, the Sobel operator, the Laplace operator, the Kirsh operator, etc. can be used, which are not specifically limited here.

In some optional implementations of this embodiment, referring to FIG. 3 , the process of frequency-dividing the original image into a high-frequency image and a medium-low frequency image may be performed through the following sub-steps S11 to S12:

In sub-step S11, a Gaussian filter operator is used to filter the original image to obtain a medium and low frequency image.

Specifically, a Gaussian filter operator (which can be denoted as mask) can be used to filter the original image (which can be regarded as a pixel matrix, denoted as I) to obtain a filtered low-frequency image (which can be regarded as a pixel matrix, denoted as I) lf). which is:

lf=I×mash

Among them, the mask can be preset, for example, can be set to:

Sub-step S12: Determine the first pixel difference between the original image and the medium and low frequency image by pixel, and use the image formed by the first pixel difference of each pixel as the high frequency image.

Specifically, taking 64×64 pixels contained in both the original image and the low-frequency image as an example, at this time, both I and lf are pixel matrices of 64×64. For each pixel point in the pixel matrix, the pixel value of the pixel point in the original image and the pixel value of the pixel point in the medium and low frequency image can be different to obtain the first pixel difference value corresponding to the pixel point. After performing the above operations on each pixel, 64×64 first pixel difference values can be obtained. The 64×64 first pixel difference values constitute a new 64×64 pixel matrix, and the pixel matrix can be used as a high-frequency image. Here, the high-frequency image can be denoted as hf, then hf=I-lf.

In step 202, frequency division processing is performed on the mid-low frequency image to obtain the mid-frequency image and the low-frequency image.

In this embodiment, the executing subject of the image processing method may first filter the mid-low frequency image to obtain the mid-frequency image. For example, edge-preserving filters such as bilateral filtering, guided filtering, or other directional filters may be used to filter the mid-low frequency image to obtain the mid-frequency image, and the filtering method used is not specifically limited here. After the intermediate frequency image is obtained, the low frequency image can be calculated based on the intermediate and low frequency image and the intermediate frequency image.

In some optional implementations of this embodiment, referring to FIG. 4 , the process of dividing the frequency of the original image into a high-frequency image and a medium-low frequency image may be performed through the following sub-steps S21 to S23:

In sub-step S21, the medium and low frequency images are filtered to obtain an intermediate frequency image (which may be denoted as mf).

Here, the filtering result may be referred to as I _F, the intermediate image can be calculated by the following equation:

mf=lf- _IF .

In sub-step S22, a target coefficient (which can be recorded as Gain) is determined based on the degree of dermabrasion set by the user.

Wherein, the value of Gain can be within a preset range (eg [1, 2]). The degree of microdermabrasion set by the user is positively correlated with Gain. That is, the higher the degree of microdermabrasion set by the user, the greater the value of Gain.

Sub-step S23: Determine the second pixel difference between the mid-low frequency image and the mid-frequency image multiplied by the target coefficient by pixel, and use the image formed by the second pixel difference of each pixel as the low-frequency image (can be denoted as dc). That is: dc=lf-Gain×mf.

By setting a target coefficient associated with the degree of microdermabrasion, users can be supported to manually adjust the degree of beautification microdermabrasion, so as to flexibly meet user needs.

In some optional implementations of this embodiment, referring to FIG. 5 , the foregoing sub-step S21 may be further specifically performed according to the following sub-steps S211 to S215:

In sub-step S211, face detection is performed on the original image to determine the face area in the original image.

The above executive body may use a pre-trained face detection model to perform face detection. The above face detection model can be obtained by pre-training the basic model based on the machine learning method. The above basic models may include but are not limited to CNN (Convolutional Neural Network, Convolutional Neural Network), R-CNN (Regions with Convolutional Neural Network Features), DCNN (Deep Convolutional Network, Deep Convolutional Neural Network), Faster-RCNN (Faster-RCNN) -Regions with Convolutional Neural Network Features), SSD (Single Shot MultiBox Detector), etc.

Sub-step S212, detecting the skin type of the face region.

The above executive body can detect the skin type of the face region in various ways. For example, a skin type detection model for detecting skin type can be pre-trained, and the face region is input into the skin type detection model, so as to obtain the skin type. The above skin type detection model can be obtained by training a machine learning method (such as a supervised learning method). The sample set for training the model can include multiple face image samples. Each face image sample can have an annotation representing the skin type.

Optionally, the above-mentioned execution subject may also detect the skin type of the face region through the following steps: first, perform face key point detection on the face region to obtain the coordinates of the face key points. Among them, the key points of the face may include the key points of the outline of the facial features of the face. Then, based on the coordinates of the key points of the face, the facial features in the face region are determined. For example, for a nose, a nose contour may be determined based on key points in the contour of the nose, and an area surrounded by the contour may be determined as a nose area. After the facial features area is determined, the face area is converted into a brightness map, and the facial features area in the brightness map is removed to obtain the remaining area. Among them, Y represents the brightness value of a pixel in the brightness map, and R, G, B represent the three-channel value of the pixel in the image of the face area, then Y=0.299×R+0.57×G+0.114 ×B. Finally, based on the remaining areas in the luminance map, the skin type of the face area is determined.

Among them, based on the remaining areas in the luminance map, the skin type of the face area can be determined by the following steps:

The first step is to filter the remaining area to obtain the gradient of each pixel in the remaining area.

Here, various filtering algorithms can be used to filter the remaining regions, such as Roberts operator, Prewitt operator, and Sobel operator. Taking the use of the Sobel operator as an example, the Sobel operator can be used to filter the remaining area first to obtain the horizontal gradient and vertical gradient of each pixel in the remaining area. Then, for each pixel in the remaining area, the sum of the absolute value of the horizontal gradient and the vertical gradient of the pixel is determined as the gradient of the pixel. Thus, the gradient of each pixel in the remaining area is obtained. For each pixel, the gradient of the pixel can be denoted as gard, then:

in,

Represents a convolution operation. The two matrices in the above formula are Sobel operators.

The second step is to perform statistics on the gradient, and determine the skin type of the face region based on the statistical results.

In some examples, pixels with a gradient smaller than a preset threshold may be used as flat skin pixels, and the ratio of the number of flat skin pixels to the total number of pixels in the remaining area may be determined; then, based on the ratio, the skin of the face area may be determined. quality category.

In other examples, the sum of the gradients of each pixel point in the remaining area may be determined; and then, based on the sum of the gradients, the skin type of the face area is determined. For example, the skin type can be denoted as skin, and if the sum of the gradients is less than th1, then skin=0. If the sum of the gradients is greater than or equal to th1 and less than or equal to th2, then skin=1. If the sum of gradients is greater than th2, then skin=2. Among them, th1 and th2 are preset thresholds, and th1 is smaller than th2. skin=0, skin=1, and skin=2 represent three skin types: smooth skin, normal skin, and scarred skin, respectively.

Sub-step S213, construct the background mask of the original image.

The above actor can construct the background mask of the original image in various ways. The background mask here can be a weight map, such as a weight map whose weight values are in the numerical interval [0, 1]. The greater the weight, the greater the probability of belonging to the face region.

As an example, the original image can be first converted to a binary image. For example, the weight value of the background area is set to 0, and the weight value of the face area is set to 1. Then, a mean filter operator is used to filter the binary image to obtain a background mask. After mean filtering, there can be a transition value at the junction of the background area and the face area, and the transition value is greater than 0 and less than 1.

As another example, the face contour point detection may be performed on the face region first to obtain the coordinates of the face contour point. Face contour points can be detected by means of face key point detection. Then, the coordinates of the face contour points can be down-sampled to obtain the down-sampled face contour point coordinates. For example, when performing double downsampling, if the original coordinates of a face contour point in the face region are (100, 100), the downsampled coordinates are (50, 50). Afterwards, a downsampled binary image can be generated based on the downsampled face contour point coordinates. Among them, the pixel value is 1 in the area surrounded by the down-sampled face contour points; the pixel value is 0 outside the area surrounded by the down-sampled face contour points. Finally, the binary image is upsampled by interpolation to obtain the background mask of the original image. Through interpolation, a transition value can exist at the junction between the background area and the face area in the background mask, and the transition value is greater than 0 and less than 1.

In sub-step S214, the filtering strength is determined based on the skin type and the background mask.

Here, a mapping table may be preset to represent the corresponding relationship between the skin type, the value of the background mask and the filtering strength. Among them, the larger the value of the background mask, the greater the filtering strength. The greater the skin quality estimation result value (that is, the worse the skin quality), the greater the filtering strength. Based on this mapping table, the filtering strength of each pixel can be viewed.

Sub-step S215, filter the medium and low frequency images based on the filtering strength to obtain an intermediate frequency image.

Here, filter() can be used to represent the filtering process, and ε represents the mapping table. Then the filtering result in the above sub-step 21 _IF =filter(lf, ε).

Since the larger the value of the background mask, the stronger the filtering strength, so this filtering method can perform relatively strong filtering on the face area. Since the larger the skin quality estimation result value is, the stronger the filtering strength is, so relatively strong filtering can be performed when the skin quality is poor. In this way, different intensities of filtering can be implemented for different regions and different skin quality levels, so as to achieve the purpose of facial microdermabrasion and beauty that preserves the contour of the face and maintains the clarity of the background. At the same time, by setting the mapping table, the filter can have the functions of edge protection and filtering strength flexibly and controllable.

Step 202: Perform intensity suppression on the pixels in the high-frequency image that satisfy the preset condition.

Step 203 , correcting the pixel points in the intermediate frequency image whose pixel values are outside the preset value range.

Step 204: Perform image fusion on the intensity-suppressed high-frequency image, the modified intermediate-frequency image, and the low-frequency image to generate a target image.

For the above steps 202 to 204 in this embodiment, reference may be made to steps 102 to 104 in the corresponding embodiment of FIG. 1 , and details are not repeated here.

In the method provided by the above-mentioned embodiments of the present application, the target coefficient Gain is used in the frequency division process, which can support the user to manually adjust the degree of skin beautification and dermabrasion, so as to flexibly meet the needs of the user. At the same time, the value of the background mask and the type of skin type are considered in the frequency division process, so that different areas and different skin quality levels can be filtered with different intensities, so as to preserve the face contour and maintain the background clarity of the face. The purpose of skin resurfacing.

Further referring to FIG. 6 , as an implementation of the methods shown in the above figures, the present application provides an embodiment of an electronic device, which corresponds to the method embodiment shown in FIG. 6 .

As shown in FIG. 6 , the electronic device described in this embodiment includes: a memory 601 and a processor 602 .

The memory 601 is used to store program instructions; the processor 602 is used to execute the program instructions stored in the memory, and when the program instructions are executed, the processor 602 is used to perform the following steps:

In some optional implementation manners of this embodiment, the performing intensity suppression on the pixels in the high-frequency image that meet the preset condition includes: sequentially taking each pixel in the high-frequency image as a target point, perform the following steps: determine the neighborhood of the target point; obtain the maximum pixel value and/or the minimum pixel value in the neighborhood; in response to the difference between the pixel value of the target point and the maximum pixel value greater than the first A preset value, or the difference between the minimum pixel value and the pixel value of the target point is greater than a second preset value, and the pixel value of the target point is corrected.

In some optional implementation manners of this embodiment, the modifying the pixel value of the target point includes: using a median filtering algorithm to filter the pixel value of the target point, to obtain the corrected pixel value of the target point. pixel value.

In some optional implementations of this embodiment, the modifying the pixel points in the intermediate frequency image whose pixel values are outside the preset value range includes: for each pixel point in the intermediate frequency image, responding to If the pixel value of the pixel point is smaller than the minimum value in the preset value range, the pixel value of the pixel point is adjusted to the minimum value.

In some optional implementations of this embodiment, the modifying the pixel points in the intermediate frequency image whose pixel values are outside the preset value range includes: for each pixel point in the intermediate frequency image, responding to If the pixel value of the pixel point is greater than the maximum value in the preset value range, the pixel value of the pixel point is adjusted to the maximum value.

In some optional implementations of this embodiment, the processor 602 further performs the following steps: performing skin color detection on the original image, and determining the skin color confidence of the pixels in the original image; The confidence level of the skin color of the pixels in the image, and the original image and the target image are image-fused to generate the final image.

In some optional implementations of this embodiment, performing image fusion on the original image and the target image based on the skin color confidence of the pixels in the original image to generate a final image, including: for For each pixel, the first pixel value of the pixel is obtained from the target image, and the second pixel value of the pixel is obtained from the original image; the skin color confidence of the pixel is taken as the first pixel value. The weight of a pixel value, the difference between the preset value and the weight is used as the weight of the second pixel value; the weighted summation is performed on the first pixel value and the second pixel value to obtain the pixel point The final pixel value of ; based on the final pixel value of each pixel, the final image is generated.

In some optional implementations of this embodiment, performing frequency division processing on the original image to obtain a high-frequency image, an intermediate-frequency image and a low-frequency image includes: performing frequency division processing on the original image to obtain a high-frequency image and middle and low frequency images; frequency division processing is performed on the middle and low frequency images to obtain an intermediate frequency image and a low frequency image.

In some optional implementation manners of this embodiment, performing frequency division processing on the original image to obtain a high-frequency image and a medium-low frequency image includes: filtering the original image with a Gaussian filter operator to obtain a medium-frequency image. Low-frequency image; determine the first pixel difference between the original image and the medium-low-frequency image by pixel, and use the image formed by the first pixel difference of each pixel as the high-frequency image.

In some optional implementations of this embodiment, the performing frequency division processing on the medium and low frequency images to obtain an intermediate frequency image and a low frequency image includes: filtering the medium and low frequency images to obtain an intermediate frequency image; Determine the target coefficient according to the set degree of microdermabrasion; determine the second pixel difference between the mid-low frequency image and the mid-frequency image multiplied by the target coefficient according to the pixel point, and determine the second pixel difference value of each pixel by the second pixel difference. The constructed image is regarded as a low frequency image.

In some optional implementation manners of this embodiment, the filtering the mid-low frequency image to obtain the mid-frequency image includes: performing face detection on the original image, and determining a face region in the original image ; Detect the skin quality category of the face area; construct the background mask of the original image, and the background mask is a weight map; Based on the skin quality category and the background mask, determine the filtering strength; The intermediate and low frequency images are filtered by the filtering strength to obtain an intermediate frequency image.

In some optional implementations of this embodiment, the detecting the skin type of the face region includes: performing face key point detection on the face region to obtain the coordinates of the face key points; based on The coordinates of the key points of the face determine the facial features area in the facial area; convert the facial area into a luminance map, remove the facial features area in the luminance graph, and obtain the remaining area; based on the The remaining area in the luminance map determines the skin type of the face area.

In some optional implementations of this embodiment, the determining the skin type of the face region based on the remaining region in the luminance map includes: filtering the remaining region to obtain the The gradient of each pixel in the remaining area is calculated; the gradient is counted, and the skin type of the face area is determined based on the statistical result.

In some optional implementation manners of this embodiment, the filtering the remaining area to obtain the gradient of each pixel in the remaining area includes: using a Sobel operator to filter the remaining area , to obtain the horizontal gradient and vertical gradient of each pixel in the remaining area; for each pixel in the remaining area, the sum of the absolute value of the horizontal gradient and the vertical gradient of the pixel is determined as the pixel. gradient.

In some optional implementations of this embodiment, the performing statistics on the gradients and determining the skin quality category of the face region based on the statistical results includes: using pixels with gradients smaller than a preset threshold as flat skins quality pixels, determine the ratio of the number of flat skin quality pixels to the total number of pixels in the remaining area; based on the ratio, determine the skin quality category of the face area.

In some optional implementations of this embodiment, the performing statistics on the gradients, and determining the skin type of the face region based on the statistical results, includes: determining a difference between the gradients of the pixels in the remaining region. and; based on the sum of the gradients, determine the skin type of the face region.

In some optional implementations of this embodiment, the constructing the background mask of the original image includes: converting the original image into a binary image; Filter processing to get the background mask.

In some optional implementations of this embodiment, the constructing the background mask of the original image includes: performing face contour point detection on the face region to obtain the coordinates of the face contour points; The coordinates of the face contour points are down-sampled to obtain the down-sampled face contour point coordinates; based on the down-sampled face contour point coordinates, a down-sampled binary image is generated; The binary image is upsampled to obtain the background mask of the original image.

The electronic device provided by the above-mentioned embodiments of the present application can suppress the blackheads on the face while retaining the high-frequency texture of the image by suppressing the intensity of the pixels satisfying the preset conditions in the high-frequency image obtained after frequency division. High frequency imperfections. By correcting the pixel points whose pixel values are outside the preset value range in the intermediate frequency image obtained after frequency division, it is possible to remove the dark parts such as facial color spots while retaining the contrast information such as the contour of the face. By image fusion of the high-frequency image after intensity suppression, the modified intermediate-frequency image, and the low-frequency image obtained after frequency division, the facial color spots, blackheads and other defects in the original image can be eliminated, and the facial texture information and The contrast feature enhances the three-dimensionality of the face in the image.

Embodiments of the present application further provide a computer-readable medium, where a computer program is stored on the computer-readable medium. When the computer program is executed by a processor, each process of the foregoing image processing method embodiments can be realized, and the same technical effect can be achieved. . In order to avoid repetition, when the computer program is executed by the processor, each process of the embodiments of the above-mentioned methods is implemented, which will not be repeated here.

The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments may be referred to each other.

It should be understood by those skilled in the art that the embodiments of the present application may be provided as a method, an apparatus, or a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-readable media having computer-usable program code embodied therein, including but not limited to disk storage, CD-ROM, optical storage, and the like.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to the present application. It will be understood that each flow and/or block in the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing terminal equipment to produce a machine that causes the instructions to be executed by the processor of the computer or other programmable data processing terminal equipment Means are created for implementing the functions specified in the flow or flows of the flowcharts and/or the blocks or blocks of the block diagrams.

These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing terminal equipment to operate in a particular manner, such that the instructions stored in the computer readable memory result in an article of manufacture comprising instruction means, the The instruction means implement the functions specified in the flow or flow of the flowcharts and/or the block or blocks of the block diagrams.

These computer program instructions can also be loaded on a computer or other programmable data processing terminal equipment, so that a series of operational steps are performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby executing on the computer or other programmable terminal equipment The instructions executed on the above provide steps for implementing the functions specified in the flowchart or blocks and/or the block or blocks of the block diagrams.

While the preferred embodiments of the present application have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once armed with the basic inventive concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiment and all changes and modifications that fall within the scope of this application.

Finally, it should also be noted that in this document, relational terms such as first and second are used only to distinguish one entity or operation from another, and do not necessarily require or imply these entities or that there is any such actual relationship or sequence between operations. Moreover, the terms "comprising", "comprising" or any other variation thereof are intended to encompass non-exclusive inclusion such that a process, method, article or terminal device that includes a list of elements includes not only those elements, but also a non-exclusive list of elements. other elements, or also include elements inherent to such a process, method, article or terminal equipment. Without further limitation, an element defined by the phrase "comprises a..." does not preclude the presence of additional identical elements in the process, method, article or terminal device comprising said element.

The data transmission system, method, sending end and computer-readable medium provided by the present application have been introduced in detail above. The principles and implementations of the present application are described with specific examples. The description of the above embodiments is only used for In order to help understand the method of the present application and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation and application scope. In summary, this specification The content should not be construed as a limitation on this application.

Claims

An image processing method, characterized in that the method comprises:

Perform frequency division processing on the original image to obtain a high-frequency image, an intermediate-frequency image and a low-frequency image;

performing intensity suppression on the pixels that meet the preset conditions in the high-frequency image;

Correcting the pixels whose pixel values are outside the preset value range in the intermediate frequency image;

Image fusion is performed on the intensity-suppressed high-frequency image, the modified intermediate-frequency image, and the low-frequency image to generate a target image.
The method according to claim 1, wherein the performing intensity suppression on the pixels satisfying a preset condition in the high-frequency image comprises:

Take each pixel in the high-frequency image as a target point in turn, and perform the following steps:

determine the neighborhood of the target point;

obtaining the maximum pixel value and/or the minimum pixel value in the neighborhood;

In response to the difference between the pixel value of the target point and the maximum pixel value being greater than the first preset value, or the difference between the minimum pixel value and the pixel value of the target point being greater than the second preset value, The pixel value of the target point is corrected.
The method according to claim 2, wherein the modifying the pixel value of the target point comprises:

A median filtering algorithm is used to filter the pixel value of the target point to obtain the corrected pixel value of the target point.
The method according to claim 1, wherein the modifying the pixels whose pixel values are outside the preset value range in the intermediate frequency image comprises:

For each pixel in the intermediate frequency image, in response to the pixel value of the pixel being less than the minimum value in the preset value range, the pixel value of the pixel is adjusted to the minimum value.
The method according to claim 1 or 4, wherein the modifying the pixels whose pixel values are outside the preset value range in the intermediate frequency image comprises:

For each pixel in the intermediate frequency image, in response to the pixel value of the pixel being greater than the maximum value in the preset value range, the pixel value of the pixel is adjusted to the maximum value.
The method according to claim 1, wherein the method further comprises:

skin color detection is performed on the original image to determine the skin color confidence of the pixels in the original image;

Based on the confidence of the skin color of the pixels in the original image, image fusion is performed on the original image and the target image to generate a final image.
The method according to claim 6, wherein, performing image fusion on the original image and the target image based on the confidence of skin color of pixels in the original image to generate a final image, comprising:

For each pixel, the first pixel value of the pixel is obtained from the target image, and the second pixel value of the pixel is obtained from the original image; the skin color confidence of the pixel is used as the The weight of the first pixel value, the difference between the preset value and the weight is used as the weight of the second pixel value; the weighted summation is performed on the first pixel value and the second pixel value to obtain the pixel value the final pixel value of the point;

Based on the final pixel value of each pixel, the final image is generated.
The method according to claim 1, wherein the performing frequency division processing on the original image to obtain a high-frequency image, an intermediate-frequency image and a low-frequency image, comprising:

Perform frequency division processing on the original image to obtain a high frequency image and a medium and low frequency image;

The intermediate and low frequency images are subjected to frequency division processing to obtain an intermediate frequency image and a low frequency image.
The method according to claim 8, wherein the performing frequency division processing on the original image to obtain a high-frequency image and a medium-low frequency image, comprising:

The original image is filtered by the Gaussian filter operator to obtain the medium and low frequency image;

Determine the first pixel difference between the original image and the medium and low frequency image by pixel, and take the image formed by the first pixel difference of each pixel as the high frequency image.
The method according to claim 8, wherein the performing frequency division processing on the medium and low frequency images to obtain the intermediate frequency images and the low frequency images, comprising:

Filtering the medium and low frequency images to obtain an intermediate frequency image;

Determine the target coefficient based on the microdermabrasion degree set by the user;

Determine the second pixel difference between the intermediate and low frequency image and the intermediate frequency image multiplied by the target coefficient according to pixel points, and use the image formed by the second pixel difference of each pixel as the low frequency image.
The method according to claim 10, wherein the filtering the mid-low frequency image to obtain the mid-frequency image comprises:

performing face detection on the original image to determine the face area in the original image;

detecting the skin type of the face region;

constructing a background mask of the original image, and the background mask is a weight map;

determining a filtering strength based on the skin type and the background mask;

The intermediate and low frequency images are filtered based on the filtering strength to obtain an intermediate frequency image.
The method according to claim 11, wherein the detecting the skin type of the face region comprises:

Performing face key point detection on the face area to obtain the coordinates of the face key points;

Determine the facial features area in the face area based on the coordinates of the key points of the face;

Converting the face area into a brightness map, removing the facial features in the brightness map, and obtaining the remaining area;

Based on the remaining regions in the luminance map, a skin quality category of the face region is determined.
The method according to claim 12, wherein the determining the skin type of the face region based on the remaining region in the luminance map comprises:

Filtering the remaining area to obtain the gradient of each pixel in the remaining area;

Statistics are performed on the gradient, and the skin type of the face region is determined based on the statistical result.
The method according to claim 13, wherein the filtering the remaining area to obtain the gradient of each pixel in the remaining area, comprising:

The remaining area is filtered by the Sobel operator to obtain the horizontal gradient and the vertical gradient of each pixel in the remaining area;

For each pixel point in the remaining area, the sum of the absolute value of the horizontal gradient and the vertical gradient of the pixel point is determined as the gradient of the pixel point.
The method according to claim 13, wherein the performing statistics on the gradients, and determining the skin type of the face region based on the statistical results, comprises:

Taking the pixels whose gradient is less than the preset threshold as the flat skin quality pixels, the ratio of the number of the flat skin quality pixels to the total number of pixels in the remaining area is determined;

Based on the ratio, the skin type of the face region is determined.
The method according to claim 13, wherein the performing statistics on the gradients, and determining the skin type of the face region based on the statistical results, comprises:

determining the sum of the gradients of the pixels in the remaining area;

Based on the sum of the gradients, the skin type of the face region is determined.
The method according to claim 11, wherein the constructing the background mask of the original image comprises:

converting the original image into a binary image;

A mean filter operator is used to filter the binary image to obtain a background mask.
The method according to claim 11, wherein the constructing the background mask of the original image comprises:

Performing face contour point detection on the face region to obtain the coordinates of the face contour points;

down-sampling the coordinates of the face contour points to obtain down-sampled face contour point coordinates;

Based on the down-sampled face contour point coordinates, a down-sampled binary image is generated;

The binary image is up-sampled by a difference method to obtain a background mask of the original image.
An electronic device, comprising: a processor and a memory;

the memory for storing program instructions;

The processor executes the program instructions stored in the memory, and when the program instructions are executed, the processor is configured to perform the following steps:

Perform frequency division processing on the original image to obtain a high-frequency image, an intermediate-frequency image and a low-frequency image;

performing intensity suppression on the pixels that meet the preset conditions in the high-frequency image;

Correcting the pixels whose pixel values are outside the preset value range in the intermediate frequency image;

Image fusion is performed on the intensity-suppressed high-frequency image, the modified intermediate-frequency image, and the low-frequency image to generate a target image.
The electronic device according to claim 19, wherein the performing intensity suppression on the pixels satisfying a preset condition in the high-frequency image comprises:

Take each pixel in the high-frequency image as a target point in turn, and perform the following steps:

determine the neighborhood of the target point;

obtaining the maximum pixel value and/or the minimum pixel value in the neighborhood;

In response to the difference between the pixel value of the target point and the maximum pixel value being greater than the first preset value, or the difference between the minimum pixel value and the pixel value of the target point being greater than the second preset value, The pixel value of the target point is corrected.
The electronic device according to claim 20, wherein the modifying the pixel value of the target point comprises:

A median filtering algorithm is used to filter the pixel value of the target point to obtain the corrected pixel value of the target point.
The electronic device according to claim 19, wherein the revising the pixels whose pixel values are outside the preset value range in the intermediate frequency image comprises:

For each pixel in the intermediate frequency image, in response to the pixel value of the pixel being less than the minimum value in the preset value range, the pixel value of the pixel is adjusted to the minimum value.
The electronic device according to claim 19, wherein the revising the pixels whose pixel values are outside the preset value range in the intermediate frequency image comprises:

For each pixel in the intermediate frequency image, in response to the pixel value of the pixel being greater than the maximum value in the preset value range, the pixel value of the pixel is adjusted to the maximum value.
The electronic device according to claim 19, wherein the processor further performs the following steps:

skin color detection is performed on the original image to determine the skin color confidence of the pixels in the original image;

Based on the confidence of the skin color of the pixels in the original image, image fusion is performed on the original image and the target image to generate a final image.
The electronic device according to claim 24, wherein the image fusion of the original image and the target image based on the confidence of the skin color of the pixels in the original image to generate a final image comprises:

For each pixel, the first pixel value of the pixel is obtained from the target image, and the second pixel value of the pixel is obtained from the original image; the skin color confidence of the pixel is used as the The weight of the first pixel value, the difference between the preset value and the weight is used as the weight of the second pixel value; the weighted summation is performed on the first pixel value and the second pixel value to obtain the pixel value the final pixel value of the point;

Based on the final pixel value of each pixel, the final image is generated.
The electronic device according to claim 19, wherein the performing frequency division processing on the original image to obtain a high-frequency image, an intermediate-frequency image and a low-frequency image, comprising:

Perform frequency division processing on the original image to obtain a high frequency image and a medium and low frequency image;

The intermediate and low frequency images are subjected to frequency division processing to obtain an intermediate frequency image and a low frequency image.
The electronic device according to claim 26, wherein the performing frequency division processing on the original image to obtain a high-frequency image and a medium-low frequency image, comprising:

The original image is filtered by the Gaussian filter operator to obtain the medium and low frequency image;

Determine the first pixel difference between the original image and the medium and low frequency image by pixel, and take the image formed by the first pixel difference of each pixel as the high frequency image.
The electronic device according to claim 26, wherein the performing frequency division processing on the mid- and low-frequency images to obtain an intermediate-frequency image and a low-frequency image, comprising:

Filtering the medium and low frequency images to obtain an intermediate frequency image;

Determine the target coefficient based on the microdermabrasion degree set by the user;

Determine the second pixel difference between the intermediate and low frequency image and the intermediate frequency image multiplied by the target coefficient according to pixel points, and use the image formed by the second pixel difference of each pixel as the low frequency image.
The electronic device according to claim 28, wherein the filtering the mid-low frequency image to obtain the mid-frequency image comprises:

performing face detection on the original image to determine the face area in the original image;

detecting the skin type of the face region;

constructing a background mask of the original image, and the background mask is a weight map;

determining a filtering strength based on the skin type and the background mask;

The intermediate and low frequency images are filtered based on the filtering strength to obtain an intermediate frequency image.
The electronic device according to claim 29, wherein the detecting the skin type of the face region comprises:

Performing face key point detection on the face area to obtain the coordinates of the face key points;

Determine the facial features area in the face area based on the coordinates of the key points of the face;

Converting the face area into a brightness map, removing the facial features in the brightness map, and obtaining the remaining area;

Based on the remaining regions in the luminance map, a skin quality category of the face region is determined.
The electronic device according to claim 30, wherein the determining the skin type of the face region based on the remaining region in the luminance map comprises:

Filtering the remaining area to obtain the gradient of each pixel in the remaining area;

Statistics are performed on the gradient, and the skin type of the face region is determined based on the statistical result.
The electronic device according to claim 31, wherein the filtering the remaining area to obtain the gradient of each pixel in the remaining area comprises:

The remaining area is filtered by the Sobel operator to obtain the horizontal gradient and the vertical gradient of each pixel in the remaining area;

For each pixel point in the remaining area, the sum of the absolute value of the horizontal gradient and the vertical gradient of the pixel point is determined as the gradient of the pixel point.
The electronic device according to claim 31, wherein the performing statistics on the gradients, and determining the skin type of the face region based on the statistical results, comprises:

Taking the pixels whose gradient is less than the preset threshold as the flat skin quality pixels, the ratio of the number of the flat skin quality pixels to the total number of pixels in the remaining area is determined;

Based on the ratio, the skin type of the face region is determined.
The electronic device according to claim 31, wherein the performing statistics on the gradients, and determining the skin type of the face region based on the statistical results, comprises:

determining the sum of the gradients of the pixels in the remaining area;

Based on the sum of the gradients, the skin type of the face region is determined.
The electronic device according to claim 29, wherein the constructing the background mask of the original image comprises:

converting the original image into a binary image;

A mean filter operator is used to filter the binary image to obtain a background mask.
The electronic device according to claim 29, wherein the constructing the background mask of the original image comprises:

Performing face contour point detection on the face area to obtain the coordinates of the face contour points;

down-sampling the coordinates of the face contour points to obtain down-sampled face contour point coordinates;

Based on the down-sampled face contour point coordinates, a down-sampled binary image is generated;

Up-sampling the binary image by means of difference value to obtain the background mask of the original image.
A computer-readable medium on which a computer program is stored, characterized in that, when the program is executed by a processor, the method according to any one of claims 1-18 is implemented.