CN107871303B - An image processing method and device - Google Patents

Abstract

Embodiments of the present invention disclose an image processing method and device. The method includes: separating luminance components and chrominance components of an original image to obtain an original luminance image and an original chrominance image; filtering the original luminance image based on a fast filtering algorithm to obtain a first luminance image; Curve adjustment is performed on the image to obtain a second brightness image; based on the alpha fusion algorithm and the first brightness image, image fusion is performed on the original brightness image and the second brightness image to obtain a target brightness image; based on the target brightness image and the original color image degree image to obtain the target image. When applying the solution provided by the embodiment of the present invention for image processing, while ensuring the processing effect, it can also ensure that the image processing algorithm occupies less system resources, and the speed of image processing is improved compared with the prior art.

Description

An image processing method and device

technical field

The present invention relates to the field of computer technology, and in particular, to an image processing method and device.

Background technique

Image processing refers to the technique of analyzing images with a computer to achieve the desired results. Image beautification processing is a common image processing method, which refers to an image processing method that uses a computer to process image information to meet human visual psychology or application needs, such as face beautification. With the rapid development of computer technology, image beautification processing is also constantly developing to meet people's needs.

In the prior art, an image beautification algorithm based on curve adjustment is a commonly used image beautification processing method. Through this algorithm, the brightness of the darker defect areas in the image can be appropriately adjusted, while the contrast of the original image can also be preserved, such as , when beautifying the face area in the image, first find the spot area in the face area of the image through this algorithm, and then perform Gaussian blurring on the entire image to realize the beautification of the face area in the image. Especially dark areas such as hair retain better contrast, and the result image obtained after processing is clearer. This algorithm can increase the brightness of the spot area in the image, so that the brightness of the spot is close to the skin color of the human body, so as to achieve the effect of face beautification.

However, in the existing image beautification algorithms based on curve adjustment, the blur radius used in Gaussian blur processing is large, which results in a large amount of system resources when the entire algorithm is running, and therefore requires high performance of the device, so that mobile devices, etc. Devices with poor performance cannot complete real-time processing of images.

SUMMARY OF THE INVENTION

The embodiment of the present invention discloses an image processing method and device, which are used for reducing the system resources occupied by the image processing algorithm in the process of beautifying the image. The technical solution is as follows:

In order to achieve the above purpose, in a first aspect, an embodiment of the present invention provides an image processing method, and the method includes:

Separate the luminance component and the chrominance component of the original image to obtain the original luminance image and the original chrominance image;

Based on a fast filtering algorithm, filtering processing is performed on the original luminance image to obtain a first luminance image;

performing curve adjustment on the original brightness image to obtain a second brightness image;

Based on the alpha fusion algorithm and the first brightness image, image fusion is performed on the original brightness image and the second brightness image to obtain a target brightness image;

Based on the target luminance image and the original chrominance image, a target image is obtained.

Preferably, the fast filtering algorithm is any one of a guided graph filtering method, a fast Gaussian filtering method and a box filtering method.

Preferably, performing image fusion on the original brightness image and the second brightness image based on an alpha fusion algorithm and the first brightness image to obtain a target brightness image, including:

According to the pixel position, the difference between the brightness components of each pixel in the original brightness image and the first brightness image is calculated respectively, and the brightness difference value of each pixel is obtained;

Determine the alpha component corresponding to each pixel point based on the brightness difference value of each pixel point;

Based on the alpha component corresponding to each pixel point, image fusion is performed on the original luminance image and the second luminance image to obtain a target luminance image.

Preferably, based on the fast filtering algorithm, the original brightness image is filtered to obtain the first brightness image, including:

performing reduction processing on the original brightness image to obtain a third brightness image;

Based on the fast filtering algorithm and the third luminance image, a first luminance image obtained by filtering the original luminance image is obtained.

Preferably, when the fast filtering algorithm is a guide map filtering method, the filtering process is performed on the original brightness image based on the fast filtering algorithm to obtain a first brightness image, including:

Perform filtering processing on each pixel in the original luminance image in the following manner, and obtain a first luminance image according to the filtering result:

determining a pixel point in a first type of preset window centered on a first pixel point, wherein the first pixel point is any pixel point in the original brightness image;

Calculate the first variance and the first average value corresponding to the brightness component of the determined pixel point;

Filter processing is performed on the first pixel point according to the first variance and the first average value.

Preferably, the calculating the first variance and the first average value corresponding to the luminance component of the determined pixel point includes:

obtaining a fourth brightness image, wherein the fourth brightness image is: an image determined according to the square value of the Y component of each pixel in the original brightness image;

obtaining a first integral image and a second integral image respectively corresponding to the original luminance image and the fourth luminance image;

Based on the first integral image and the second integral image, a first variance and a first average value corresponding to the luminance components of the determined pixel points are calculated.

Preferably, performing filtering processing on the first pixel point according to the first variance and the first average value includes:

According to the following formula, filter processing is performed on the first pixel point:

为所述第一平均值。In the formula, Y _b is the filtered luminance component of the first pixel; σ ₁ is the first variance; ε ₁ is the preset first smoothing parameter; Y is the luminance component of the first pixel ;

is the first average value.

Preferably, the curve adjustment is performed on the original brightness image to obtain a second brightness image, including:

According to the following formula, adjust the luminance component of each pixel in the original luminance image to obtain a second luminance image:

In the formula, k is the scale coefficient, Y _h (i, j) represents the adjusted brightness component corresponding to the pixel point whose pixel coordinate is (i, j) in the original brightness image, and Y(i, j) represents the The luminance component of the pixel whose pixel coordinate is (i, j) in the original luminance image.

Preferably, the obtaining the target image based on the target luminance image and the original chrominance image includes:

Perform guide map filtering processing on the original chromaticity image to obtain a target chromaticity image;

The target image is obtained by combining the target luminance image and the target chrominance image.

Preferably, the original chrominance image includes a first chrominance image corresponding to the U component and a second chrominance image corresponding to the V component,

The performing guide map filtering processing on the original chromaticity image to obtain a target chromaticity image, including:

In the following manner, filtering processing is performed on each pixel in the first chromaticity image, and a first target chromaticity image is obtained according to the filtering result, and filtering processing is performed on each pixel in the second chromaticity image, And obtain the second target chrominance image according to the filtering result:

Determine the pixels in the second type of preset window centered on the second pixel, where the second pixel is any pixel in the first chromaticity image or the second chromaticity image ;

Calculate the second variance and the second average value corresponding to the pixel value of the determined pixel point;

Filter processing is performed on the second pixel point according to the second variance and the second average value.

Preferably, performing filtering processing on the second pixel points according to the second variance and the second average value includes:

According to the following formula, filter processing is performed on the second pixel point:

为所述第二平均值。In the formula, X _b is the filtered pixel value of the second pixel point; σ ₂ is the second variance; ε ₂ is the preset second smoothing parameter; X is the pixel value of the second pixel point;

is the second average value.

In a second aspect, an embodiment of the present invention provides an image processing apparatus, and the apparatus includes:

The separation module is used to separate the luminance component and the chrominance component of the original image to obtain the original luminance image and the original chrominance image;

a filtering processing module, configured to perform filtering processing on the original luminance image based on a fast filtering algorithm to obtain a first luminance image;

a curve adjustment module, configured to perform curve adjustment on the original brightness image to obtain a second brightness image;

a fusion module, configured to perform image fusion on the original brightness image and the second brightness image based on an alpha fusion algorithm and the first brightness image to obtain a target brightness image;

An obtaining module is configured to obtain a target image based on the target luminance image and the original chrominance image.

Preferably, the fast filtering algorithm is any one of a guided graph filtering method, a fast Gaussian filtering method and a box filtering method.

Preferably, the fusion module includes:

a first calculation sub-module, configured to calculate the difference between the brightness components of each pixel in the original brightness image and the first brightness image according to the pixel position, and obtain the brightness difference value of each pixel;

The first determination sub-module is used to determine the alpha component corresponding to each pixel point based on the brightness difference value of each pixel point;

The fusion sub-module is configured to perform image fusion on the original brightness image and the second brightness image based on the alpha component corresponding to each pixel to obtain a target brightness image.

Preferably, the filtering processing module includes:

a processing submodule, used for reducing the original brightness image to obtain a third brightness image;

The obtaining sub-module is configured to obtain a first luminance image obtained by filtering the original luminance image based on the fast filtering algorithm and the third luminance image.

Preferably, when the fast filtering algorithm is a directed graph filtering method, the filtering processing module is specifically used for:

Perform filtering processing on each pixel in the original brightness image, and obtain a first brightness image according to the filtering result;

Wherein, the filtering processing module includes:

a second determination sub-module, configured to determine a pixel point in the first type of preset window centered on the first pixel point; the first pixel point is any pixel point in the original brightness image;

The second calculation submodule is used to calculate the first variance and the first average value corresponding to the brightness component of the determined pixel point;

A filtering processing sub-module, configured to perform filtering processing on the first pixel point according to the first variance and the first average value.

Preferably, the second calculation submodule includes:

a first obtaining unit, configured to obtain a fourth brightness image, wherein the fourth brightness image is: an image determined according to the square value of the Y component of each pixel in the original brightness image;

a second obtaining unit, configured to obtain a first integral image and a second integral image respectively corresponding to the original luminance image and the fourth luminance image;

A first calculation unit, configured to calculate a first variance and a first average value corresponding to the luminance component of the determined pixel point based on the first integral image and the second integral image.

Preferably, the filtering processing sub-module is specifically used for:

According to the following formula, filter processing is performed on the first pixel point:

为所述第一平均值。In the formula, Y _b is the filtered luminance component of the first pixel; σ ₁ is the first variance; ε ₁ is the preset first smoothing parameter; Y is the luminance component of the first pixel ;

is the first average value.

Preferably, the curve adjustment module is specifically used for:

According to the following formula, adjust the luminance component of each pixel in the original luminance image to obtain a second luminance image:

In the formula, k is the scale coefficient, Y _h (i, j) represents the adjusted brightness component corresponding to the pixel point whose pixel coordinate is (i, j) in the original brightness image, and Y(i, j) represents the The luminance component of the pixel whose pixel coordinate is (i, j) in the original luminance image.

Preferably, the obtaining module includes:

A guide map filtering processing sub-module, configured to perform guide map filtering processing on the original chromaticity image to obtain a target chromaticity image;

The merging submodule is used for merging the target luminance image and the target chrominance image to obtain a target image.

Preferably, the original chrominance image includes: a first chrominance image corresponding to the U component and a second chrominance image corresponding to the V component, and the guide map filtering processing sub-module is specifically used for:

Perform filtering processing on each pixel in the first chromaticity image, and obtain a first target chromaticity image according to the filtering result, perform filtering processing on each pixel in the second chromaticity image, and obtain a first target chromaticity image according to the filtering result. obtain a second target chromaticity image;

Wherein, the guide map filtering processing sub-module includes:

A determination unit, configured to determine a pixel point within a second type of preset window centered on a second pixel point, wherein the second pixel point is within the first chromaticity image or the second chromaticity image any pixel of ;

a second calculation unit, configured to calculate the second variance and the second average value corresponding to the pixel value of the determined pixel point;

A filtering processing unit, configured to perform filtering processing on the second pixel point according to the second variance and the second average value.

Preferably, the filtering processing unit is specifically used for:

According to the following formula, filter processing is performed on the second pixel point:

为所述第二平均值。In the formula, X _b is the filtered pixel value of the second pixel point; σ ₂ is the second variance; ε ₂ is the preset second smoothing parameter; X is the pixel value of the second pixel point;

is the second average value.

It can be seen from the above that in the solution provided by the embodiment of the present invention, the luminance component and the chrominance component of the original image are first separated to obtain the original luminance image and the original chrominance image; and then the original luminance image is filtered based on a fast filtering algorithm. , obtain a first brightness image; and perform curve adjustment on the original brightness image to obtain a second brightness image; then based on the alpha fusion algorithm and the first brightness image, perform image fusion on the original brightness image and the second brightness image to obtain A target luminance image; finally, a target image is obtained based on the target luminance image and the original chrominance image. Compared with the prior art, the filtering algorithm adopted in this solution is a fast filtering algorithm which is different from the Gaussian filtering method. When such a fast filtering algorithm performs filtering processing on an image, it can ensure the image processing effect while ensuring the image processing effect. The algorithm occupies less system resources and the image processing speed is fast. Therefore, this solution can also be applied to devices with poor performance such as mobile devices, so that the devices with poor performance can process high-quality video images in real time.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative efforts.

FIG. 1 is a first schematic flowchart of an image processing method provided by an embodiment of the present invention;

2 is a function image of formula (1) involved in an image processing method provided by an embodiment of the present invention;

3 is a function image of a function f(x) involved in an image processing method provided by an embodiment of the present invention;

4 is a second schematic flowchart of an image processing method provided by an embodiment of the present invention;

FIG. 5 is a first structural schematic diagram of an image processing apparatus according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of a second structure of an image processing apparatus according to an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

The following briefly introduces the technical terms involved in the present invention.

Image Processing: As mentioned above, image processing techniques refer to techniques that use computers to analyze images to achieve desired results. In the prior art, image processing can be performed on a single image captured by industrial cameras, scanners, and other equipment. Of course, each frame of image captured by a camera or played by a video player when playing a video can also be processed. Image processing is performed in real time.

Luminance component and chrominance component of an image: In the prior art, YUV is a color coding method mainly used in TV systems and analog video fields. A YUV image is divided into three components. Luma), which is the grayscale value; and "U" and "V" represent the chroma (Chrominance or Chroma), which is used to describe the color and saturation of the image, and is used to specify the color of the pixel. The luminance component of the image may refer to the Y component of the image in the YUV encoding format, and the chrominance component may refer to the U component and the V component of the image in the YUV encoding format.

The present invention will be described in detail below through specific embodiments.

FIG. 1 is a first schematic flowchart of an image processing method provided by an embodiment of the present invention, and the method includes:

S101: Separate the luminance component and the chrominance component of the original image to obtain the original luminance image and the original chrominance image.

The luminance component here corresponds to the Y component, and the chrominance component may refer to the U component and the V component. As mentioned above, the image in this color space includes three components, so the luminance component and the chrominance component here can be separated according to the prior art.

For example, for an image in I420 format commonly used in digital cameras, it is well known to those skilled in the art that when an image in I420 format is stored, the Y component of each pixel in the image is first stored, then the U component of each pixel is stored, and finally The V component of each pixel; when reading the data in the image, find the starting points of the three components of Y, U and V, respectively, and then read all three Y, U and V from the image. Components; it can be seen that for images in I420 format, the separation of luminance components and chrominance components is very convenient and fast, and no additional memory copying and color conversion are required.

It can be understood that in the obtained original luminance image, the luminance component of each pixel is the luminance component corresponding to the pixel at the same position in the original image. Similarly, in the obtained original chromaticity image, the color of each pixel is The chrominance component is the chrominance component corresponding to the pixel at the same position in the original image.

S102: Based on a fast filtering algorithm, perform filtering processing on the original luminance image to obtain a first luminance image.

The fast filtering algorithm here is any one of the guided graph filtering method, the fast Gaussian filtering method and the box filtering method. The above-mentioned guide map filtering method, fast Gaussian filtering method and box filtering method are commonly used filtering algorithms in the prior art, and those skilled in the art can refer to the prior art for image memory filtering processing, which will not be repeated in the present invention.

Compared with the existing Gaussian filtering method applied to image processing, this kind of fast filtering algorithm can ensure that the image processing algorithm occupies less system resources without reducing the image processing effect when filtering the image. The image processing speed is fast. For example, when such a fast filtering algorithm runs, the CPU (Central Processing Unit, central processing unit) occupancy rate of the device is low, and the heat generation of the device is reduced.

In the embodiment of the present invention, in order to speed up the filtering processing, the above-mentioned filtering processing is performed on the original brightness image based on the fast filtering algorithm to obtain the first brightness image (S102), which may include two steps:

The first step: reducing the original brightness image to obtain a third brightness image.

It can be understood that when the image is reduced, the image may be reduced by a preset multiple, and of course, the image may also be reduced to a preset size. In the third luminance image obtained after reduction, the luminance component of each pixel point may be an average value of luminance components of a plurality of corresponding pixel points in the original luminance image.

For example, if the length and width of the original brightness image are reduced by 4 times, then in the third brightness image obtained at this time, the brightness component corresponding to the pixel whose coordinate value is (1, 1) should be: the coordinate in the original brightness image The average value of the luminance components of all pixels in the rectangular area composed of two pixels whose values are (1, 1) and (4, 4).

Step 2: Based on the fast filtering algorithm and the third luminance image, obtain a first luminance image obtained by filtering the original luminance image.

The number of pixels in the third brightness image is obviously smaller than the number of pixels in the original brightness image, so after the downscaling process, the number of pixels that need to be filtered is reduced, thereby speeding up the image filtering process.

It can be understood that in the second step described here, a fast filtering algorithm needs to be performed on the third luminance image, and in addition, the image obtained after executing the filtering algorithm should also be enlarged to obtain the first luminance image. For a specific amplification processing method, reference may be made to the prior art, and detailed descriptions are not provided in the embodiment of the present invention.

It should be noted that, compared with box blur and fast Gaussian filtering, the guided graph filtering method occupies less system resources. Therefore, in this embodiment of the present invention, the fast filtering algorithm is preferably the guided graph filtering method. When the fast filtering algorithm is a guide map filtering method, the above-mentioned filtering processing is performed on the original brightness image based on the fast filtering algorithm to obtain a first brightness image (S102), including:

Perform filtering processing on each pixel in the original luminance image in the following manner, and obtain a first luminance image according to the filtering result:

Step 1: Determine a pixel point in the first type of preset window centered on the first pixel point, where the first pixel point is any pixel point in the original brightness image.

The first type of preset window refers to a window with a preset window size value, the center of which is the above-mentioned first pixel point, that is, the pixel point currently undergoing filtering processing. For example, when filtering the pixel point a in the original brightness image, the first pixel point is the pixel point a; assuming that the size of the first type of preset window is 5 × 5, the pixel coordinate of the pixel point a is ( x, y), then the first type of preset window corresponding to the pixel point a is: in the original brightness image, the pixel coordinates are two of (x-2, y-2) and (x+2, y+2) respectively The pixel points are used as a rectangular area formed by the upper left corner and the lower right corner. At this time, the determined pixel points are all the pixel points in the rectangular area.

In actual use, the window size value of the first type of preset window should be selected according to the size of the image. For example, for a 1920×1080 image, the window size value can be selected as 5×5.

Step 2: Calculate the first variance and the first average value corresponding to the luminance component of the determined pixel point.

After the pixel points in the first type of preset window are determined, the first variance and the first average value can be obtained by calculation according to a general mathematical formula. However, when the value of the window size of the first type of preset window is relatively large, the computation time for calculating the first average value and the first variance will be longer, for example, for a 11×11 type 1 preset window window, in which the total number of pixel points is 121, so the amount of data used to calculate the first average value and the first variance is large, so that the operation time is long.

In this embodiment of the present invention, in order to speed up the calculation of the first variance and the first average value, an integral graph may be used to speed up the processing. The first variance and the first average value corresponding to the brightness component of the determined pixel point in the above calculation include:

Step 1: Obtain a fourth brightness image, where the fourth brightness image is an image determined according to the square value of the Y component of each pixel in the original brightness image.

That is to say, in the fourth brightness image, the brightness component of each pixel is the square of the brightness component of the pixel corresponding to the same position in the original brightness image. For example, in the original brightness image, the coordinate value is (m, n) The luminance component of the pixel point b is M, then in the fourth luminance image, the luminance component of the pixel point b' whose coordinate value is (m, n) is M ² .

Step 2: obtaining a first integral image and a second integral image respectively corresponding to the original luminance image and the fourth luminance image;

It can be understood that the luminance component of any pixel (x, y) of the integral image is equal to the sum of the luminance components of all pixels in the rectangular area formed by the pixel from the upper left corner of the original image. Therefore, in the first integral image, the luminance component of any pixel (x, y) is equal to the sum of all the pixels in the rectangular area formed from the upper left corner of the original luminance image to the pixel (x, y) of the original luminance image. The sum of luminance components; in the second integral image, the luminance component of any pixel (x, y) is equal to the rectangular area formed from the upper left corner of the fourth luminance image to the pixel (x, y) of the fourth luminance image The sum of the luminance components of all pixels within.

Step 3: Based on the first integral image and the second integral image, calculate the first variance and the first average value corresponding to the luminance component of the determined pixel point.

在第二积分图像中，以坐标为(100，50)的像素点为中心的11×11区域内，所有像素点的亮度分量的平均值为

则此时所确定的像素点对应的第一方差

For example, the coordinates of the first pixel are (100, 50), and the window size of the first preset window is 11×11. It is assumed that in the original brightness image, the pixel with coordinates (100, 50) is the center In the 11×11 area of the average value of the luminance components of all pixels is

In the second integral image, in the 11×11 area centered on the pixel with coordinates (100, 50), the average value of the luminance components of all pixels is

Then the first variance corresponding to the pixel point determined at this time

Wherein, it is further assumed that in the first integral image, the luminance component of the pixel with coordinates (100, 50) is I ₁ , the luminance component of the pixel with coordinates (105, 44) is I ₂ , and the coordinates are (94, 55) the brightness component of the pixel point is I ₃ , and the brightness component of the pixel point whose coordinates are (94, 44) is I ₄ , then the first average value corresponding to the determined pixel point

It is well known to those skilled in the art that the integral graph can be used to accelerate the calculation of variance and average value, and the specific principle may refer to the prior art, which is not described in detail in this embodiment of the present invention.

The third step: filtering the first pixel point according to the first variance and the first average value.

After obtaining the first variance and the first average value corresponding to the first pixel, those skilled in the art can obtain the filtered value corresponding to the luminance component of the filtered first pixel based on the formula in the prior art. However, in the embodiment of the present invention, the filtering processing performed on the first pixel point according to the first variance and the first average value may be:

The first pixel is filtered according to the following formula:

为该第一平均值。In the formula, Y _b is the luminance component filtered by the first pixel point, that is, the luminance component of the corresponding pixel point in the first luminance image; σ ₁ is the first variance; ε ₁ is the preset first smoothing parameter; Y is the luminance component of the first pixel;

is the first average.

In the actual application process, the value of ε ₁ can be based on the actual situation, and is usually 15-20.

Compared with the filtering processing formula adopted by the fast guide map filtering method in the prior art, the filtering processing formula adopted in the present invention is simpler, the operation of averaging the coefficients twice is removed, and the calculation filter is reduced. The cumbersomeness of the value, and almost does not affect the processing effect of the image.

It should be noted that, in the embodiment of the present invention, optimally, the filtering processing operation may be performed in combination with the above-mentioned reduction processing technology and integral graph acceleration processing technology.

For example, reduce the original brightness image by k times, and denote the image obtained after the reduction as X; then calculate the square value of the Y component of each pixel in the image X to obtain the image Y; then obtain the corresponding image X and the image Y respectively The integral image W and integral image V of .

那么此时，在原始亮度图像中以(ki，kj)为左上角，大小为k×k的正方形区域内任一像素点都按照如下公式计算滤波后的值：Then, based on the integral image W and the integral image V, in the image X, the variance and the average value corresponding to all the pixel points in the window with the preset window size value corresponding to each pixel point are calculated. For example, in the window corresponding to the pixel whose pixel coordinate is (i, j) in the image X, the variance of all the pixels is σ ₃ , and the average value is

Then, in the original brightness image, with (ki, kj) as the upper left corner, any pixel in a square area of size k×k calculates the filtered value according to the following formula:

In the formula, Y' is the luminance component of a certain pixel in the square area in the original luminance image, Y _c is the filtered luminance component corresponding to the pixel in the original luminance image, and ε ₁ is still the aforementioned first smoothing parameter.

S103: Perform curve adjustment on the original brightness image to obtain a second brightness image.

The curve adjustment of the original luminance image can be realized by the convex function in the prior art, for example, the adjustment by a cubic spline curve. In order to make the parameter adjustment of curve adjustment more convenient, a brand-new quadratic curve is set, that is, in the embodiment of the present invention, the above-mentioned curve adjustment is performed on the original brightness image to obtain a second brightness image (S103), including:

According to the following formula, adjust the luminance component of each pixel in the original luminance image to obtain a second luminance image:

In the formula, k is the scale coefficient, Y _h (i, j) represents the adjusted brightness component corresponding to the pixel whose pixel coordinate is (i, j) in the original brightness image, and Y(i, j) represents the original brightness The luminance component of the pixel whose pixel coordinate is (i, j) in the image.

It can be understood that when the value of k is large, the increase of the brightness component after the curve adjustment may be too large, and when the value of k is small, the increase of the brightness component after the curve adjustment may be small, so In practical applications, preferably, k=0.4-0.6.

The formula for this curve adjustment can be obtained by combining the following two formulas:

c(x)=k(1-x)x+x (1)

将输入的值转换为[0，1]中的一个数值。The function image corresponding to formula (1) is shown in Figure 2. Since the value range and definition domain of the function corresponding to formula (1) are both [0, 1], it must pass

Convert the entered value to a numeric value in [0, 1].

It should be emphasized that in the above curve adjustment formula, the value of 255 is determined based on the commonly used 256-level grayscale. In actual use, 1024-level grayscale may also appear, so the above curve adjustment formula can be for:

In the formula, t represents the number of gray levels. For example, for 1024-level grayscale, t=1024.

It can be understood that in the second brightness image obtained by using the above curve adjustment formula, the brightness component of each pixel point is greater than the brightness component of the pixel point at the same position in the original brightness image, so the second brightness image can be regarded as a A highlight image of the original luminance image. In addition, for an image containing a human face, since the brightness of the image is improved, it is equivalent to whitening the skin color of the human face in the image. Therefore, compared with the prior art, when the embodiment of the present invention is used for image beautification, it does not require Extra whitening operation.

It should be emphasized that the curve adjustment is not limited to increasing the brightness component, but also reducing the brightness component. For example, using a concave function to reduce the brightness component, in the obtained second brightness image, the brightness component of each pixel is smaller than the original brightness component. The luminance component of the pixel at the same position in the luminance image.

S104: Based on the alpha fusion algorithm and the first brightness image, perform image fusion on the original brightness image and the second brightness image to obtain a target brightness image.

In this embodiment of the present invention, based on the alpha fusion algorithm and the first brightness image, image fusion is performed on the original brightness image and the second brightness image to obtain a target brightness image (S104), which may include the following three step:

Step 1: Calculate the difference between the luminance components of each pixel in the original luminance image and the first luminance image according to the pixel position, and obtain the luminance difference value of each pixel.

It should be noted that the difference obtained by directly calculating the difference between the luminance components of the same pixel in the original luminance image and the first luminance image, taking 256-level grayscale as an example, the difference will be in the range of -255 to 255. In order to To control the range from 0 to 255, first add an offset of 128 to the difference by the following formula:

D(i,j)=clip(Y(i,j) _-Yb (i,j)+128);

In the formula, D(i,j) represents the difference value obtained initially, Y(i,j) represents the brightness component of the pixel whose coordinates are (i,j) in the original brightness image, and _Yb (i,j) represents the first The luminance component of a pixel whose coordinates are (i, j) in a luminance image. In addition, clip() represents a function defined as follows:

In this embodiment of the present invention, the initially obtained difference value may be used as the brightness difference value of the pixel point finally obtained in step 1. However, in order to increase the difference range of the brightness difference value, the initially obtained difference value can be input into the following function as an independent variable:

The function image of the function f(x) is shown in Figure 3. The value obtained by multiplying the result value calculated by the function f(x) by the value 255 can be used as the brightness difference value of the pixel point finally obtained in step 1. Of course, multiple stretching operations can also be performed through f(x), that is, to further increase the difference magnitude of the luminance difference value. The embodiment of the present invention is not limited to using f(x) to complete the stretching operation, and may also be a linear transformation in the prior art.

The specific stretching times can be determined according to the actual situation. Usually, three stretching can be carried out. At this time, the brightness difference value of the pixel points finally obtained is:

In the formula, M(i,j) is the brightness difference value of the pixel point finally obtained.

It should be noted that, taking M(i, j) as the luminance component of the pixel point (i, j), a luminance difference image can be obtained. In addition, each pixel involved in this step may refer to each pixel in the original brightness image, and it can be understood that each pixel in the original brightness image corresponds to a brightness difference value.

Step 2: Determine the alpha component corresponding to each pixel point based on the luminance difference value of each pixel point.

其中，t表示灰度级数；以256级灰度为例，此原始亮度图像中的像素点(i,j)的阿尔法分量可以为

The alpha channel value of the pixel (i, j) in the original luminance image is M(i, j), and the alpha component of the pixel (i, j) can be

Among them, t represents the number of gray levels; taking 256 levels of gray as an example, the alpha component of the pixel (i, j) in the original brightness image can be

Step 3: Based on the alpha component corresponding to each pixel point, perform image fusion on the original brightness image and the second brightness image to obtain a target brightness image.

For step 3, the following formula can be used for fusion to obtain the target brightness image:

_Yr (i,j)=(M(i,j)Y( _i ,j)+(255-M(i,j))Yh(i,j))/255;

In the formula, Y _r (i, j) represents the brightness component of the pixel whose pixel coordinate is (i, j) in the target brightness image, and Y(i, j) represents the pixel whose coordinate is (i, j) in the original brightness image The luminance component of the point, Y _h (i, j) represents the luminance component of the pixel whose pixel coordinate is (i, j) in the second luminance image.

It should be noted that step 3 can be understood as taking the above-mentioned brightness difference map as an alpha channel, and performing image fusion on the original brightness image and the second brightness image to obtain a target brightness image.

S105: Based on the target luminance image and the original chrominance image, obtain a target image.

Here, the obtained target luminance image and the original chrominance image can be directly merged to obtain the target image. The specific merging method belongs to the prior art, and is not described in detail in this embodiment of the present invention.

After applying the solution provided by the embodiment of the present invention, the speed of image processing is greatly increased, and therefore, the real-time processing of images can be easily processed on devices with poor performance such as mobile devices. Table 1 below compares the number of image frames processed per second when the prior art and the image processing method provided by the embodiment of the present invention are respectively applied to different mobile devices.

Table 1

It should be noted that the prior art in Table 1 and the image processing method of the embodiment of the present invention are not combined with the Neon (a single instruction multiple data extension structure) optimization technology in the prior art. If combined with the Neon optimization technology, the existing The technology and the image processing method provided by the embodiments of the present invention can also increase the number of image frames processed per second by 30% to 50%.

It can be seen from the above that in the solution provided by the embodiment of the present invention, the luminance component and the chrominance component of the original image are first separated to obtain the original luminance image and the original chrominance image; and then the original luminance image is filtered based on a fast filtering algorithm. , obtain a first brightness image; and perform curve adjustment on the original brightness image to obtain a second brightness image; then based on the alpha fusion algorithm and the first brightness image, perform image fusion on the original brightness image and the second brightness image to obtain A target luminance image; finally, a target image is obtained based on the target luminance image and the original chrominance image. Compared with the prior art, the filtering algorithm adopted in this solution is a fast filtering algorithm which is different from the Gaussian filtering method. When such a fast filtering algorithm performs filtering processing on an image, it can ensure the image processing effect while ensuring the image processing effect. The algorithm occupies less system resources and the image processing speed is fast. Therefore, this solution can also be applied to devices with poor performance such as mobile devices, so that the devices with poor performance can process high-quality video images in real time.

Taking the face beautification algorithm in the image as an example, the embodiment of the present invention can beautify the face in the image, so that the face in the image is whiter and smoother, and achieves the effect of microdermabrasion. The embodiments of the present invention can be applied to a software product with a beauty function; and can also be used in a software product with a face detection function. After the software product with a face detection function detects a human face in an image, the The face is beautified.

In the field of image face beautification technology, it is well known that with the popularity of mobile phones, there are more and more applications of mobile phone cameras. Among them, the applications related to mobile phone photography are the most and the demand is also the largest. However, traditional mobile phones can only simply take pictures, and the beautification of pictures generally needs to be done manually on a PC (personal computer, personal computer). Although there are some automated tools, most of these tools are designed for the CPU of the PC. Recently, with the increasing demand for photo beautification, some apps (Applications) for photo beautification have appeared on mobile phones, but most of these apps are designed to beautify a single photo after taking a photo. Few can achieve real-time beauty processing during preview.

In the prior art, the beautification algorithm based on curve adjustment can be divided into the following steps: first, use Gaussian filtering to find dark areas such as spots, and use a large Gaussian radius to perform Gaussian filtering on the brightness map of the original image, then Subtract the Gaussian filtered image from the original image brightness image to get a difference image. The dark area in the difference image is the area where the spots are located; after finding the dark area, adjust the brightness of the original image by using a The convex curve remaps the brightness of the original image, so that a highlight image is obtained. After appropriate adjustment, the brightness of the spots in the highlight image can be close to the brightness of the skin color in the original image, and then the difference map is mapped to [ 0,1] interval, and use this as the alpha (alpha) channel to fuse the original image with the highlight image, so as to obtain an image with spots removed or darkened, so as to achieve the effect of beauty.

The brightness component of the beautifying algorithm image based on curve adjustment is adjusted according to a smooth curve as a whole, so the contrast can be better preserved for particularly dark areas such as hair, and the entire image obtained after image adjustment is clearer. However, in the prior art, when the beautifying algorithm based on curve adjustment performs image processing, it is necessary to first find the spot area, perform Gaussian blur on the image, and select a Gaussian blur with a relatively large blur radius. It takes up a lot of system resources, and the algorithm runs slowly. For large video images, it cannot be processed in real time on devices with poor performance such as mobile phones.

By applying the embodiments of the present invention, while ensuring the image processing effect, it can also ensure that the image processing algorithm occupies less system resources, and the image processing speed is fast. When processing video images with a resolution of 1280×720 on a mobile device, the mobile device only needs a single-core CPU Real-time image processing can be achieved, the CPU usage of mobile devices is low, and the heat generation is low.

It should be noted that there are currently two main categories of beauty algorithms for mobile devices: one is a beauty algorithm based on curve adjustment, and the other is a beauty algorithm based on edge-preserving filtering.

The beautification algorithm based on edge-preserving filtering can be subdivided into multiple categories according to the selected edge-preserving filtering algorithm. Commonly used edge-preserving filters include bilateral filtering, surface blurring, local mean square error, guided map filtering, and field transformation. The basic idea of this kind of beauty algorithm is to use edge-preserving filtering to process the original image, so as to smooth out the spots and acne on the face, while for the eyes and eyebrows, the area is relatively large, and the contrast area is basically maintained. unchanged, so as to achieve the effect of face beautification.

The beautification algorithm based on edge-preserving filtering can well remove the spots and acne on the face of the image, but it is easy to blur the thin objects such as hair in the image, and the edge of the nose in the image is easy to be blurred. The places where the equal divisions are not particularly obvious are easy to cause blurring; although the areas with obvious contrast such as eyes in the image can be maintained well, there is still a certain degree of blurring, and the image as a whole will have a certain hazy feeling.

However, when the directed graph filtering method is selected as the fast filtering algorithm in the embodiment of the present invention, there is no problem that the image definition is reduced due to the edge-preserving filtering algorithm in the prior art.

With respect to FIG. 1 , as shown in the second schematic flow chart of an image processing method shown in FIG. 4 , the above-mentioned obtaining the target image based on the target luminance image and the original chromaticity image (S105) may include:

S1051: Perform guide map filtering processing on the original chrominance image to obtain a target chrominance image.

It should be noted that the original chrominance image includes the first chrominance image corresponding to the U component and the second chrominance image corresponding to the V component. In this embodiment of the present invention, in addition to filtering the original luminance image, the The chrominance image is filtered, that is, the first chrominance image and the second chrominance image are filtered separately.

In this embodiment of the present invention, the filtering processing method for the original chrominance image may be similar to the filtering processing method for the original luminance image, and the specific details may refer to the foregoing method embodiments. Specifically, performing the guide map filtering process on the original chrominance image above to obtain the target chrominance image (S1051) may include:

Filter each pixel in the original chrominance image in the following manner, and obtain the target chrominance image according to the filtering result:

Step 1: Determine the pixel points in the second type of preset window centered on the second pixel point, wherein the second pixel point is the first chromaticity image or the second chromaticity image. any pixel.

The second type of preset window refers to a window with a preset window size value, and the center of the second type of preset window is the above-mentioned second pixel, that is, the pixel currently undergoing filtering processing. For example, when filtering the pixel point a in the first chromaticity image, the second pixel point is the pixel point a; at the same time, assuming that the size of the second type of preset window is 3×3, the size of the pixel point a is If the pixel coordinates are (x, y), the second type of preset window corresponding to the pixel point a is: in the original chromaticity image, the pixel coordinates are (x-1, y-1) and (x+1, y+ respectively) 1) The rectangular area formed by the two pixel points, at this time, the determined pixel points are all the pixel points in the rectangular area.

In actual use, the window size value of the second type of preset window should be selected according to the size of the image. For example, for a 1920×1080 image, the window size value can be selected as 3×3.

It should be noted that the window size value of the first type of preset windows corresponding to the same original image is larger than the window size value of the second type of preset windows.

Step 2: Calculate the second variance and the second average value corresponding to the pixel value of the determined pixel point. It can be understood that when the original chromaticity image is the aforementioned first chromaticity image, the pixel value is the U corresponding to the pixel point. component, when the original chromaticity image is the aforementioned second chromaticity image, the pixel value is the V component corresponding to the pixel point.

After the pixel points are determined, the second variance and the second average value can be obtained by calculation according to a general mathematical formula. However, when the window size value of the second type of preset window is larger, the calculation time of the second average value and the second variance will be longer, for example, for a 11×11 second type of preset window , the total number of pixel points is 121, so the amount of data used to calculate the second average value and the second variance is large, so the operation time is long.

In this embodiment of the present invention, in order to speed up the calculation of the second variance and the second average value, an integral graph may be used to speed up the processing. The second variance and the second average value corresponding to the pixel value of the determined pixel point in the above calculation include:

Step 1: Calculate the square value of the pixel value of each pixel in the original chromaticity image to obtain a third chromaticity image.

That is to say, in the third chromaticity image, the pixel value of each pixel is the square of the pixel value of the pixel corresponding to the same position in the original chromaticity image. For example, in the original chromaticity image, the coordinate value is (m , n) the pixel value of the pixel point b is M, then in the third chromaticity image, the pixel value of the pixel point b' with the coordinate value (m, n) is M ² .

Step 2: obtaining a third integral image and a fourth integral image respectively corresponding to the original chromaticity image and the third chromaticity image;

It can be understood that in the third integral image, the pixel value of any pixel point (x, y) is equal to all the pixels in the rectangular area from the upper left corner of the original chromaticity image to the pixel point (x, y) of the original chromaticity image. In the fourth integral image, the pixel value of any pixel (x, y) is equal to the pixel value from the upper left corner of the third chromaticity image to the pixel (x, y) of the third chromaticity image ) is the sum of the pixel values of all the pixels in the rectangular area.

Step 3: Based on the third integral image and the fourth integral image, calculate the second variance and the second average value corresponding to the pixel value of the determined pixel point.

在第四积分图像中，以坐标为(100，50)的像素点为中心的11×11区域内，所有像素点的像素值的平均值为

则此时所确定的像素点对应的第二方差

For example, the coordinates of the second pixel point above are (100, 50), and the window size of the second preset window is 11×11. Suppose that in the original chromaticity image, the pixel point with coordinates (100, 50) is In the 11×11 area in the center, the average value of all pixel points is

In the fourth integral image, in the 11×11 area centered on the pixel point whose coordinates are (100, 50), the average value of the pixel values of all the pixel points is

Then the second variance corresponding to the pixel point determined at this time

Among them, it is further assumed that in the third integral image, the pixel value of the pixel with coordinates (100, 50) is Z ₁ , the pixel value of the pixel with coordinates (105, 44) is Z ₂ , and the coordinates are (94, The pixel value of the pixel point of 55) is Z ₃ , and the pixel value of the pixel point whose coordinates are (94, 44) is Z ₄ , then the second average value corresponding to the determined pixel point

It is known to those skilled in the art that the integral map can be used to speed up the calculation of variance and average value. For the specific principle, reference may be made to the prior art and the aforementioned filtering process for the luminance component, which is not described in detail in this embodiment of the present invention.

The third step: filtering the second pixel point according to the second variance and the second average value.

In the embodiment of the present invention, the filtering processing performed on the second pixel point according to the second variance and the second average value may be:

The second pixel is filtered according to the following formula:

为该第二平均值。In the formula, X _b is the filtered pixel value of the second pixel point; σ ₂ is the second variance; ε ₂ is the preset second smoothing parameter; X is the pixel value of the second pixel point;

is the second average.

In the actual application process, the value of ε ₂ can be based on the actual situation, but corresponding to the same original image, the value of ε ₂ should be less than the value of ε ₁ , the value of ε ₂ can be 6 to 10, the optimal value is 8.

In the embodiment of the present invention, in order to make the filtering processing on the second pixel point simpler and occupy less system resources, the above filtering processing formula is used. Certainly, the formula used for the filtering processing of the second pixel point may also refer to the prior art, which is not limited in this embodiment of the present invention.

In addition, when filtering the U and V channels (chrominance components) separately, you can refer to the filtering processing of the luminance component, and perform the filtering processing of the chrominance component by combining the image reduction processing technology and the integral image acceleration processing technology.

It can be understood that filtering the U and V channels can make the color transition of the final target image more natural. Moreover, it should be emphasized that if the filtering processing of the luminance component occupies more system resources, the resources in the system may not be enough to support the filtering processing of the U and V channels (chrominance components), but due to the implementation of the present invention In an example, the filtering processing of the luminance component occupies few system resources, so there are sufficient resources in the system to support the filtering processing performed on the U and V channels (chrominance components).

S1052: Combine the target luminance image and the target chrominance image to obtain a target image.

Of course, reference is made to the prior art for a specific method of combining the target luminance image and the target chrominance image, which is not described in detail in this embodiment of the present invention.

Compared with the prior art, the filtering algorithm adopted in this solution is a fast filtering algorithm which is different from the Gaussian filtering method. When such a fast filtering algorithm performs filtering processing on an image, it can ensure the image processing effect while ensuring the image processing effect. The algorithm occupies less system resources and the image processing speed is fast. Therefore, this solution can also be applied to devices with poor performance such as mobile devices, so that the devices with poor performance can process high-quality video images in real time. More importantly, since in the embodiment of the present invention, the filtering processing of the luminance component occupies less system resources, there are sufficient resources in the system to support the filtering processing of the U and V channels (chrominance components), so that the final result is obtained. The color transition of the target image is more natural, improving the image processing effect.

Corresponding to FIG. 1 , as shown in FIG. 5 , a first structural schematic diagram of an image processing apparatus provided by an embodiment of the present invention, the apparatus includes:

The separation module 110 is used to separate the luminance component and the chrominance component of the original image to obtain the original luminance image and the original chrominance image;

A filtering processing module 120, configured to perform filtering processing on the original luminance image based on a fast filtering algorithm to obtain a first luminance image;

a curve adjustment module 130, configured to perform curve adjustment on the original brightness image to obtain a second brightness image;

A fusion module 140, configured to perform image fusion on the original brightness image and the second brightness image based on an alpha fusion algorithm and the first brightness image to obtain a target brightness image;

The obtaining module 150 is configured to obtain a target image based on the target luminance image and the original chrominance image.

Specifically, the fast filtering algorithm is any one of a guided graph filtering method, a fast Gaussian filtering method and a box filtering method.

Specifically, the fusion module 140 includes a first calculation sub-module, a first determination sub-module and a fusion sub-module (not shown in the figure).

The first calculation sub-module is configured to calculate the difference between the luminance components of each pixel in the original luminance image and the first luminance image according to the pixel position, and obtain the luminance difference value of each pixel;

The first determination submodule is used to determine the alpha component corresponding to each pixel point based on the brightness difference value of each pixel point;

The fusion sub-module is configured to perform image fusion on the original brightness image and the second brightness image based on the alpha component corresponding to each pixel to obtain a target brightness image.

Specifically, the filtering processing module 120 includes a processing sub-module and an obtaining sub-module (not shown in the figure).

The processing submodule is used for reducing the original brightness image to obtain a third brightness image;

The obtaining sub-module is configured to obtain a first luminance image obtained by filtering the original luminance image based on the fast filtering algorithm and the third luminance image.

Specifically, when the fast filtering algorithm is a directed graph filtering method, the filtering processing module 120 is specifically used for:

Perform filtering processing on each pixel in the original brightness image, and obtain a first brightness image according to the filtering result;

Wherein, the filtering processing module 120 includes a second determining sub-module, a second calculating sub-module and a filtering processing sub-module (not shown in the figure).

The second determination submodule is used to determine a pixel point in the first type of preset window centered on the first pixel point; the first pixel point is any pixel point in the original brightness image;

The second calculation submodule is used to calculate the first variance and the first average value corresponding to the brightness component of the determined pixel point;

The filtering processing sub-module is configured to perform filtering processing on the first pixel point according to the first variance and the first average value.

In practical applications, specifically, the second calculation sub-module includes a first obtaining unit, a second obtaining unit and a first calculating unit (not shown in the figure).

The first obtaining unit is configured to obtain a fourth luminance image, wherein the fourth luminance image is: an image determined according to the square value of the Y component of each pixel in the original luminance image;

the second obtaining unit, configured to obtain a first integral image and a second integral image respectively corresponding to the original luminance image and the fourth luminance image;

The first calculation unit is configured to calculate, based on the first integral image and the second integral image, a first variance and a first average value corresponding to the luminance component of the determined pixel point.

In practical applications, the filtering processing sub-module is specifically used for:

According to the following formula, filter processing is performed on the first pixel point:

为所述第一平均值。In the formula, Y _b is the filtered luminance component of the first pixel; σ ₁ is the first variance; ε ₁ is the preset first smoothing parameter; Y is the luminance component of the first pixel ;

is the first average value.

In practical applications, the curve adjustment module 130 is specifically used for:

According to the following formula, adjust the luminance component of each pixel in the original luminance image to obtain a second luminance image:

In the formula, k is the scale coefficient, Y _h (i, j) represents the adjusted brightness component corresponding to the pixel point whose pixel coordinate is (i, j) in the original brightness image, and Y(i, j) represents the The luminance component of the pixel whose pixel coordinate is (i, j) in the original luminance image.

It can be seen from the above that in the solution provided by the embodiment of the present invention, the luminance component and the chrominance component of the original image are first separated to obtain the original luminance image and the original chrominance image; and then the original luminance image is filtered based on a fast filtering algorithm. , obtain a first brightness image; and perform curve adjustment on the original brightness image to obtain a second brightness image; then based on the alpha fusion algorithm and the first brightness image, perform image fusion on the original brightness image and the second brightness image to obtain A target luminance image; finally, a target image is obtained based on the target luminance image and the original chrominance image. Compared with the prior art, the filtering algorithm adopted in this solution is a fast filtering algorithm which is different from the Gaussian filtering method. When such a fast filtering algorithm performs filtering processing on an image, it can ensure the image processing effect while ensuring the image processing effect. The algorithm occupies less system resources and the image processing speed is fast. Therefore, this solution can also be applied to devices with poor performance such as mobile devices, so that the devices with poor performance can process high-quality video images in real time.

Corresponding to the method embodiment shown in FIG. 4 , as shown in FIG. 6 , a second schematic structural diagram of an image processing apparatus provided by an embodiment of the present invention, the obtaining module 150 includes:

The guide map filtering processing sub-module 1501 is used to perform guide map filtering processing on the original chromaticity image to obtain a target chromaticity image;

The merging sub-module 1502 is used for merging the target luminance image and the target chrominance image to obtain a target image.

In practical applications, the original chrominance image includes a first chrominance image corresponding to the U component and a second chrominance image corresponding to the V component, and the guide map filtering processing sub-module is specifically used for:

Perform filtering processing on each pixel in the first chromaticity image, and obtain a first target chromaticity image according to the filtering result, perform filtering processing on each pixel in the second chromaticity image, and obtain a first target chromaticity image according to the filtering result. obtain a second target chromaticity image;

Wherein, the guide map filtering processing sub-module 1501 includes a determining unit, a third computing unit and a filtering processing unit (not shown in the figure).

A determination unit, configured to determine a pixel point within a second type of preset window centered on a second pixel point, wherein the second pixel point is within the first chromaticity image or the second chromaticity image any pixel of ;

a second calculation unit, configured to calculate the second variance and the second average value corresponding to the pixel value of the determined pixel point;

A filtering processing unit, configured to perform filtering processing on the second pixel point according to the second variance and the second average value.

Specifically, the third calculation unit may include a first calculation subunit, an obtaining subunit, and a second calculation subunit (not shown in the figure).

a first calculation subunit, used to calculate the square value of the pixel value of each pixel in the original chromaticity image to obtain a third chromaticity image;

an obtaining subunit for obtaining a third integral image and a fourth integral image corresponding to the original chromaticity image and the third chromaticity image respectively;

The second calculation subunit is configured to calculate the second variance and the second average value corresponding to the pixel value of the determined pixel point based on the third integral image and the fourth integral image.

In practical applications, the filtering processing unit is specifically used for:

According to the following formula, filter processing is performed on the second pixel point:

为所述第二平均值。In the formula, X _b is the filtered pixel value of the second pixel point; σ ₂ is the second variance; ε ₂ is the preset second smoothing parameter; X is the pixel value of the second pixel point;

is the second average value.

Compared with the prior art, the filtering algorithm adopted in this solution is a fast filtering algorithm which is different from the Gaussian filtering method. When such a fast filtering algorithm performs filtering processing on an image, it can ensure the image processing effect while ensuring the image processing effect. The algorithm occupies less system resources and the image processing speed is fast. Therefore, this solution can also be applied to devices with poor performance such as mobile devices, so that the devices with poor performance can process high-quality video images in real time. More importantly, since in the embodiment of the present invention, the filtering processing of the luminance component occupies less system resources, there are sufficient resources in the system to support the filtering processing of the U and V channels (chrominance components), so that the final result is obtained. The color transition of the target image is more natural, improving the image processing effect.

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

Each embodiment in this specification is described in a related manner, and the same and similar parts between the various embodiments may be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the apparatus embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and reference may be made to some descriptions of the method embodiments for related parts.

Those of ordinary skill in the art can understand that all or part of the steps in the implementation of the above method can be completed by instructing the relevant hardware through a program, and the program can be stored in a computer-readable storage medium. Storage medium, such as: ROM/RAM, magnetic disk, optical disk, etc.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims (20)

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

separating out a brightness component and a chrominance component of an original image to obtain an original brightness image and an original chrominance image;

based on a rapid filtering algorithm, filtering the original brightness image to obtain a first brightness image;

performing curve adjustment on the original brightness image to obtain a second brightness image;

performing image fusion on the original brightness image and the second brightness image based on an alpha fusion algorithm and the first brightness image to obtain a target brightness image;

obtaining a target image based on the target brightness image and the original chrominance image;

wherein the image fusing the original luminance image and the second luminance image based on an alpha fusion algorithm and the first luminance image to obtain a target luminance image includes:

according to the positions of the pixels, respectively calculating the difference of the brightness components of each pixel in the original brightness image and the first brightness image to obtain the brightness difference value of each pixel;

determining alpha components corresponding to the pixel points based on the brightness difference values of the pixel points;

and carrying out image fusion on the original brightness image and the second brightness image based on the alpha component corresponding to each pixel point to obtain a target brightness image.

2. The method of claim 1, wherein the fast filtering algorithm is any one of a guide map filtering method, a fast gaussian filtering method, and a box filtering method.

3. The method according to claim 1, wherein the filtering the original luminance image based on the fast filtering algorithm to obtain a first luminance image comprises:

carrying out reduction processing on the original brightness image to obtain a third brightness image;

and obtaining a first brightness image obtained by filtering the original brightness image based on a rapid filtering algorithm and the third brightness image.

4. The method according to claim 1, wherein when the fast filtering algorithm is a guide map filtering method, the filtering the original luminance image based on the fast filtering algorithm to obtain a first luminance image comprises:

according to the following mode, filtering processing is carried out on each pixel point in the original brightness image, and a first brightness image is obtained according to a filtering result:

determining pixel points in a first-class preset window with a first pixel point as a center, wherein the first pixel point is any pixel point in the original brightness image;

calculating a first variance and a first average value corresponding to the brightness component of the determined pixel point;

and carrying out filtering processing on the first pixel point according to the first variance and the first average value.

5. The method of claim 4, wherein the calculating a first variance and a first mean corresponding to the luminance component of the determined pixel point comprises:

obtaining a fourth luminance image, wherein the fourth luminance image is: determining an image according to a square value of Y component of each pixel point in the original brightness image;

obtaining a first integral image and a second integral image respectively corresponding to the original brightness image and the fourth brightness image;

and calculating a first variance and a first average value corresponding to the brightness component of the determined pixel point based on the first integral image and the second integral image.

6. The method according to claim 4, wherein the filtering the first pixel according to the first variance and the first mean value comprises:

and carrying out filtering processing on the first pixel point according to the following formula:

in the formula, Y_bFiltering the brightness component of the first pixel point; sigma₁Is the first variance;₁is a preset first smoothing parameter; y is the brightness component of the first pixel point;

is the first average value.

7. The method of claim 1, wherein the curve adjusting the original luminance image to obtain a second luminance image comprises:

adjusting the brightness component of each pixel point in the original brightness image according to the following formula to obtain a second brightness image:

wherein k is a proportionality coefficient and Y is_h(i, j) represents the adjusted brightness component corresponding to the pixel point with the pixel coordinate (i, j) in the original brightness image, and Y (i, j) represents the brightness component of the pixel point with the pixel coordinate (i, j) in the original brightness image.

8. The method of claim 1, wherein deriving a target image based on the target luma image and the original chroma images comprises:

conducting guide map filtering processing on the original chrominance image to obtain a target chrominance image;

and combining the target brightness image and the target chromaticity image to obtain a target image.

9. The method of claim 8, wherein the original chroma image comprises: a first chrominance image corresponding to the U component and a second chrominance image corresponding to the V component,

the step of performing guide map filtering processing on the original chrominance image to obtain a target chrominance image comprises:

according to the following mode, filtering each pixel point in the first chrominance image, obtaining a first target chrominance image according to a filtering result, filtering each pixel point in the second chrominance image, and obtaining a second target chrominance image according to the filtering result:

determining a pixel point in a second-class preset window with a second pixel point as a center, wherein the second pixel point is any one pixel point in the first chrominance image or the second chrominance image;

calculating a second variance and a second average value corresponding to the pixel value of the determined pixel point;

and carrying out filtering processing on the second pixel point according to the second variance and the second average value.

10. The method according to claim 9, wherein the filtering the second pixel according to the second variance and the second average value includes:

and carrying out filtering processing on the second pixel point according to the following formula:

in the formula, X_bFiltering the pixel value of the second pixel point; sigma₂Is the second variance;₂the second smoothing parameter is preset; x is the pixel value of the second pixel point;

is the second average value.

11. An image processing apparatus, characterized in that the apparatus comprises:

the separation module is used for separating the brightness component and the chrominance component of the original image to obtain an original brightness image and an original chrominance image;

the filtering processing module is used for carrying out filtering processing on the original brightness image based on a rapid filtering algorithm to obtain a first brightness image;

the curve adjusting module is used for adjusting the curve of the original brightness image to obtain a second brightness image;

the fusion module is used for carrying out image fusion on the original brightness image and the second brightness image based on an alpha fusion algorithm and the first brightness image to obtain a target brightness image;

an obtaining module, configured to obtain a target image based on the target luminance image and the original chrominance image;

wherein, the fusion module comprises:

the first calculation submodule is used for respectively calculating the difference of the brightness components of each pixel point in the original brightness image and the first brightness image according to the position of the pixel point to obtain the brightness difference value of each pixel point;

the first determining submodule is used for determining alpha components corresponding to the pixels based on the brightness difference value of each pixel;

and the fusion submodule is used for carrying out image fusion on the original brightness image and the second brightness image based on the alpha component corresponding to each pixel point to obtain a target brightness image.

12. The apparatus of claim 11, wherein the fast filtering algorithm is any one of a guide map filtering method, a fast gaussian filtering method, and a box filtering method.

13. The apparatus of claim 11, wherein the filter processing module comprises:

the processing submodule is used for carrying out reduction processing on the original brightness image to obtain a third brightness image;

and the obtaining submodule is used for obtaining a first brightness image obtained after filtering the original brightness image based on a rapid filtering algorithm and the third brightness image.

14. The apparatus according to claim 11, wherein when the fast filtering algorithm is a guided graph filtering method, the filtering processing module is specifically configured to:

filtering each pixel point in the original brightness image, and obtaining a first brightness image according to a filtering result;

wherein, the filtering processing module comprises:

the second determining submodule is used for determining pixel points in a first-class preset window with the first pixel points as centers; the first pixel point is any pixel point in the original brightness image;

the second calculation submodule is used for calculating a first variance and a first average value corresponding to the brightness component of the determined pixel point;

and the filtering processing submodule is used for carrying out filtering processing on the first pixel point according to the first variance and the first average value.

15. The apparatus of claim 14, wherein the second computation submodule comprises:

a first obtaining unit, configured to obtain a fourth luminance image, where the fourth luminance image is: determining an image according to a square value of Y component of each pixel point in the original brightness image;

a second obtaining unit configured to obtain a first integral image and a second integral image that respectively correspond to the original luminance image and the fourth luminance image;

and the first calculation unit is used for calculating a first variance and a first average value corresponding to the brightness component of the determined pixel point based on the first integral image and the second integral image.

16. The apparatus according to claim 14, wherein the filtering processing sub-module is specifically configured to:

and carrying out filtering processing on the first pixel point according to the following formula:

in the formula, Y_bFiltering the brightness component of the first pixel point; sigma₁Is the first variance;₁is a preset first smoothing parameter; y is the brightness component of the first pixel point;

is the first average value.

17. The apparatus of claim 11, wherein the curve adjustment module is specifically configured to:

adjusting the brightness component of each pixel point in the original brightness image according to the following formula to obtain a second brightness image:

wherein k is a proportionality coefficient and Y is_h(i, j) represents the adjusted brightness component corresponding to the pixel point with the pixel coordinate (i, j) in the original brightness image, and Y (i, j) represents the brightness component of the pixel point with the pixel coordinate (i, j) in the original brightness image.

18. The apparatus of claim 11, wherein the obtaining module comprises:

the guide map filtering processing submodule is used for carrying out guide map filtering processing on the original chrominance image to obtain a target chrominance image;

and the merging submodule is used for merging the target brightness image and the target chrominance image to obtain a target image.

19. The apparatus of claim 18, wherein the original chroma image comprises: the first chrominance image corresponding to the U component and the second chrominance image corresponding to the V component, the guide map filtering processing sub-module is specifically configured to:

filtering each pixel point in the first chrominance image, obtaining a first target chrominance image according to a filtering result, filtering each pixel point in the second chrominance image, and obtaining a second target chrominance image according to the filtering result;

wherein, the guide graph filtering processing submodule comprises:

the determining unit is used for determining pixel points in a second-class preset window with second pixel points as centers, wherein the second pixel points are any pixel points in the first chrominance image or the second chrominance image;

the second calculation unit is used for calculating a second variance and a second average value corresponding to the pixel value of the determined pixel point;

and the filtering processing unit is used for carrying out filtering processing on the second pixel point according to the second variance and the second average value.

20. The apparatus according to claim 19, wherein the filtering processing unit is specifically configured to:

and carrying out filtering processing on the second pixel point according to the following formula:

in the formula, X_bFiltering the pixel value of the second pixel point; sigma₂Is the second variance;₂the second smoothing parameter is preset; x is the pixel value of the second pixel point;

is the second average value.

Images