WO2019200689A1 - Image graying method and device, and storage medium - Google Patents

Abstract

An image graying method and a storage medium. Downsampling processing is performed on an input color image to obtain the energy of all color channels of a downsampled scaled image, and optimal coefficients of the color channels are obtained according to an iterative algorithm of transform coefficients so as to perform grey level transformation on an original color image according to the optimal coefficients to obtain a gray level image. According to the image graying method, downsampling processing is firstly performed on the color image, so that the color image reduces the redundancy of image information while retaining primary image information, thereby facilitating reducing the computational complexity in the process of obtaining the optimal coefficients in the later period; according to the image graying method, an iterative algorithm is introduced to obtain the optimal coefficients of the color channels, so that there is a good matching degree between the transform coefficients of the color channels and the color image, loss of the image information in the color image graying process can be effectively reduced, the primary image information of the original color image can be retained to the greatest extent, and the fidelity of the gray level image is facilitated to be improved.

Description

Image graying method and device, storage medium Technical field

The present invention relates to the field of image processing technologies, and in particular, to an image graying method and apparatus, and a storage medium.

Background technique

Image grayscale processing is a common method when processing images, and mainly refers to a process of converting a color image into a grayscale image. The RGB color coding method is a common form of color image. The color of each pixel in the image is determined by three components: R, G, and B. Each component has 255 values, so that one pixel can have more than 1600. The range of variation of 10,000 (255*255*255) colors, and the grayscale image is a special color image when the three components of R, G, and B are the same, and each pixel can have a variation range of 256 colors. Therefore, the grayscale image still reflects the distribution characteristics of the whole image in terms of chromaticity and brightness as the color image. In digital image processing, the images of various formats are generally converted into grayscale images to reduce the calculation amount of subsequent images. .

Image grayscale processing is suitable for a wide range of fields, such as printers in daily life, which need to print color digital documents into grayscale paper documents to save costs, and the quality of image grayscale processing will directly affect the documents. Print effect. In addition, in daily image and video algorithms, in order to speed up the operation of image grayscale processing and save hardware memory and computing resources, a good image graying algorithm is often needed to improve the real-time performance of the processing process. Fidelity, while ensuring low power consumption and real-time image grayscale processing capability, it is also necessary to avoid the loss of image information during the grayscale processing of color images.

Now the general grayscale algorithm is to perform a linear floating-point multiplication and addition operation on the basis of the intensity values of the R, G, and B channels to obtain a grayscale image. This method does not highlight the image well. The original content, especially if the overall color of the original image content is relatively close. In addition, some people (such as the image processing technology team of the Chinese University of Hong Kong) use the contrast relationship between the transformed grayscale image and the color difference image to construct a quadratic paradigm-based method to optimize the grayscale solution process. The original image information is retained to a certain extent, which is superior to the conventional linear transformation method. However, this method also has the disadvantage of complicated calculation process.

Summary of the invention

The technical problem mainly solved by the invention is how to reduce the computational complexity of the image graying algorithm, and at the same time reduce the amount of information loss in the graying process to improve the fidelity of the grayscale image.

According to a first aspect, an embodiment provides an image grayscale method, comprising the steps of:

Enter a color image;

Obtaining energy of each color channel of the color image separately;

Obtaining an optimum coefficient of each color channel according to the energy of each color channel;

Obtaining a grayscale image corresponding to the color image according to an optimal coefficient of each color channel.

The acquiring the energy of each color channel of the color image separately includes:

Performing downsampling processing on the color image to obtain a scaled image;

Obtaining an image color distribution of the scaled image, and a probability histogram of each color channel;

The energy corresponding to each color channel is obtained according to the image color distribution and the probability histogram of each color channel.

Performing a downsampling process on the color image to obtain a zoomed image, including:

Obtaining a width pixel w and a height pixel h of the color image;

When h is smaller than the first value and w is smaller than the second value, the color image is taken as the zoomed image;

When h is greater than or equal to the first value and less than or equal to the third value, or w is greater than or equal to the second value and less than or equal to the fourth value, w and h are each reduced to 1/2 of the original, and the scaled image is used as the Scale the image;

When h is greater than the third value or w is greater than the fourth value, w and h are each reduced to the original 1/4, and the scaled image is taken as the scaled image.

The obtaining an image color distribution according to the zoomed image includes:

Performing cluster analysis on the scaled image to represent similar colors on the scaled image with the same color;

And quantizing the scaled image into a color class diagram according to a result of cluster analysis;

The image color distribution is obtained from the color class diagram.

The obtaining the optimal coefficients of the respective color channels according to the energy of the respective color channels, including:

Energy and coefficient acquisition step: comparing energy of each color channel to obtain minimum energy; respectively obtaining transform coefficients of each color channel, and using a transform coefficient of a color channel corresponding to the minimum energy as a maximum transform coefficient;

Iterative step: updating the transform coefficients of the respective color channels according to the maximum transform coefficient and the preset adjustment step size and adjustment range;

The first determining step: determining whether the preset number of iterations is reached, and if not, repeating the iterative step, and vice versa, outputting the latest transform coefficient of each color channel as the optimal coefficient of the corresponding channel.

The iterative steps include:

Self-adjusting step: the maximum transform coefficient is automatically reduced by one adjustment step, and the transform coefficient of one color channel adjacent to the color channel having the largest transform coefficient is incremented by one adjustment step, and the sum of the transform coefficients according to all color channels is 1 Then, the transform coefficients of the remaining color channels are adjusted; the energy of each color channel is calculated and updated according to the adjusted transform coefficients of the respective color channels, and the energy after updating any color channel is determined to be smaller than the minimum of each color channel obtained before the update. When energy is used, the energy of the color channel is updated to the minimum energy obtained before the update;

a second determining step: determining whether the sum of the adjustment step lengths of the maximum transform coefficients is up to a preset adjustment amount, and if not, repeating the self-adjusting step, and vice versa, outputting the latest adjusted transform coefficients of the respective color channels And compare to get the maximum transform coefficient.

The second determining step includes:

Obtaining an adjustment range according to the maximum transform coefficient and the preset adjustment amount;

Determining whether the adjusted maximum transform coefficient exceeds the adjustment range;

If not, the self-adjusting step is repeated until the maximum transform coefficient is about to exceed the adjustment range. Otherwise, the newly adjusted transform coefficients of the respective color channels are outputted, and the maximum transform coefficients are compared.

And obtaining, according to an optimal coefficient of each color channel, a grayscale image corresponding to the color image, including:

The grayscale image corresponding to the color image is calculated according to the optimal coefficient of each color channel and the brightness corresponding to each color channel.

According to a second aspect, an embodiment provides an image grayscale device, including:

An input unit for inputting a color image;

a color channel acquiring unit, configured to respectively acquire energy of each color channel of the color image;

An optimal coefficient acquisition unit, configured to obtain an optimal coefficient of each color channel according to the energy of each color channel;

And a calculating unit, configured to obtain a grayscale image corresponding to the color image according to an optimal coefficient of each color channel.

The image gradation device further includes a display unit for displaying the grayscale image and/or the color image.

According to a third aspect, an embodiment provides a computer readable storage medium comprising a program executable by a processor to implement the method of the first aspect described above.

The beneficial effects of the application are:

According to the image graying method and device and the storage medium of the above embodiment, the input color image is downsampled, and the energy of each color channel of the downsampled processed image is obtained, and the iterative algorithm according to the transform coefficient is obtained. The optimum coefficient of each color channel, thereby performing gradation transformation on the original color image according to the optimal coefficients to obtain a grayscale image. Since the image is grayed out, the color image is first downsampled, so that the color image reduces the redundancy of the image information while retaining the main image information, which is beneficial to reduce the calculation amount of the process of obtaining the optimal coefficient later. Since the image graying method introduces an iterative algorithm to obtain the optimal coefficient of the color channel, the matching coefficient between the color coefficient and the color image has a good matching degree, which can effectively reduce the gray image coloring process. The loss of image information in the middle can maximize the main image information of the original color image, which is beneficial to improve the fidelity of the gray image.

DRAWINGS

1 is a flow chart of an image graying method;

2 is a flow chart of a method for obtaining channel energy;

Figure 3 is a flow chart for obtaining an image color distribution;

4 is a flow chart of an optimal coefficient iterative algorithm;

Fig. 5 is a structural diagram of an image gradation device.

detailed description

The present invention will be further described in detail below with reference to the accompanying drawings. Similar elements in different embodiments employ associated similar component numbers. In the following embodiments, many of the details are described in order to provide a better understanding of the application. However, those skilled in the art can easily realize that some of the features may be omitted in different situations, or may be replaced by other components, materials, and methods. In some cases, some operations related to the present application are not shown or described in the specification, in order to avoid that the core portion of the present application is overwhelmed by excessive description, and those skilled in the art will describe these in detail. Related operations are not necessary, they can fully understand the relevant operations according to the description in the manual and the general technical knowledge in the field.

In addition, the features, operations, or characteristics described in the specification may be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can also be sequentially changed or adjusted in a manner that can be apparent to those skilled in the art. Therefore, the various sequences in the specification and the drawings are only for the purpose of describing a particular embodiment, and are not intended to

The serial numbers themselves for the components herein, such as "first", "second", etc., are only used to distinguish the described objects, and do not have any order or technical meaning. As used herein, "connected" or "coupled", unless otherwise specified, includes both direct and indirect connections (joining).

Referring to FIG. 1, the present application discloses an image grayscale method, which includes steps S100-S400, which are respectively described below.

In step S100, a color image is input. In order to meet the application needs of different occasions, technicians have developed a variety of image color coding methods, commonly known as RGB, CMY, CMYK, HSV, HIS, YUV, each color coding method has one or more color channels, each Each color channel stores information about the color elements in the image, and the colors in all the color channels are superimposed to produce the color of the pixels in the image. Therefore, the color images input here include, but are not limited to, RGB images (RGB images include three color channels of red R, green G, and blue B).

In step S200, the energy of each color channel of the color image is respectively obtained, where the energy refers to the information cross entropy of the intensity distribution of the intensity values between the color channels after the color channels are converted according to the corresponding coefficients. In an embodiment, in order to clearly understand the process of acquiring the energy of each color channel of the color image, the embodiment uses the RGB color image as an example to illustrate how to obtain energy on the three color channels of R, G, and B. Step S200 Steps S210-S230 may be included, which are specifically described below.

In step S210, in order to reduce the redundancy information in the image and obtain a better image size, the RGB color image needs to be downsampled to obtain a zoomed image. The specific process is:

(1) Acquiring the height pixel h and the width pixel w of the RGB color image, the method of acquiring the pixel value is a commonly used method for processing an image, and belongs to the prior art, and will not be described in detail herein.

(2) Reasonably scaling the RGB color image according to the height pixel h and the width pixel w. Here, a technical means is provided, specifically: when h is smaller than the first value and w is smaller than the second value, the RGB color image is used as a zoom Image; when h is greater than or equal to the first value and less than or equal to the third value, or w is greater than or equal to the second value and less than or equal to the fourth value, w and h are each reduced to the original 1/2, and the scaled image is taken as The image is scaled; when h is greater than the third value or w is greater than the fourth value, w and h are each reduced to the original 1/4, and the scaled image is taken as the scaled image. In a specific embodiment, the first value, the second value, the third value, and the fourth value herein represent pixel reference values set by the user as needed, and are preferably set to 288, 352, 600, and 800, respectively. .

Step S220, performing color and probability statistics on the scaled image obtained in step S210 to obtain an image color distribution X of the scaled image, and a probability histogram Y _{j of} each color channel (subscript j may take 1, 2, and 3, Represents the color channels corresponding to R, G, and B).

The process of acquiring the image color distribution X may specifically include steps S221-S223, which are respectively described below.

Step S221, the scaled image is often a 24-bit bitmap, consisting of three components of R, G, and B, each component occupies 8 bits. At this time, each component is compressed from 8 bits to 5 bits, and then cluster analysis is performed ( Cluster analysis is an aggregation processing method for data, which is to aggregate similar data into a class and then perform a unified representation to represent the similar colors on the respective color channels of the scaled image in the same color, and finally make the whole frame. Images can be converted to 256 colors for uniform representation.

In a specific embodiment, the K-means algorithm can be used for clustering analysis, which belongs to the hard clustering algorithm and is representative of a typical prototype-based objective function clustering method. It is a certain distance from the data point to the prototype. As an objective function of optimization, the algorithm usually adopts the error square sum criterion function as the clustering criterion function, and uses the function to obtain the extreme value method to obtain the adjustment rules of the iterative operation to achieve the goal of the smallest evaluation index. It belongs to the prior art and will not be specifically described here.

Step S222, after performing cluster analysis on the scaled image, the entire image can be converted into 256 types of colors for unified representation. According to the result of the cluster analysis, the color class diagram color(i) is obtained, where i represents the color class. The number ranges from 0 to 255.

Step S223, the color class diagram color(i) includes the main color information of the zoomed image, and the color information is statistically obtained to obtain the image color distribution X, and the color distribution map PD of the zoomed image is obtained according to the image color distribution X ( X, i), specifically expressed as

The color distribution map PD(X, i) has a small color error between the original compressed image and does not affect the main image information in the original compressed image.

Wherein, the probability histogram distribution Y _j (including Y ₁ , Y ₂ , Y ₃ ) can take the values on the color channels of R, G, and B respectively, and obtain the probability histogram corresponding to each probability histogram according to the value obtained. PD(Y _j , i), specifically expressed as

Step S230, according to the image color distribution X and its corresponding color distribution map PD(X, i), and each probability histogram distribution Y _j and its corresponding probability histogram PD(Y _j , i) The energy corresponding to the color channel. The energy corresponding to the R color channel can be expressed as

Then, those skilled in the art can obtain the energies En(Y ₂ ) and En(Y ₃ ) of the G color channel and the B color channel, respectively, according to the calculation method.

In step S300, the optimal coefficients of the respective color channels are obtained according to the energy of each color channel. In an embodiment, as shown in FIG. 4, the R color channel, the G color channel, and the B are respectively obtained according to the energies En(Y ₁ ), En(Y ₂ ), and En(Y ₃ ) of the three color channels obtained in step S230. The optimal coefficient of the color channel may include steps S310-350, as described below.

Step S310, which may be referred to as an energy and coefficient acquisition step. In an embodiment, the energy of each color channel is compared to obtain a minimum energy, and at the same time, the transform coefficients corresponding to the three color channels R, G, and B are obtained, and the minimum energy is obtained. The transform coefficient of the corresponding color channel is taken as the maximum transform coefficient. In a specific embodiment, the three color channels R, G, and B respectively correspond to a set of transform coefficients (c_r, c_g, c_b), and in the initial stage, the initial transform coefficient corresponding to the R color channel is (1, 0, 0). The initial transform coefficient corresponding to the G color channel is (0, 1, 0), and the initial transform coefficient corresponding to the B color channel is (0, 0, 1). Then, if the energy of the R color channel is the smallest, then (1, 0, 0) is used as the reference coefficient group, where 1 can be used as the maximum transform coefficient c_max.

Step S320, which may be referred to as an iterative step. In an embodiment, the transform coefficients of the respective color channels are updated according to the maximum transform coefficient c_max and the preset adjustment step size λ and the adjustment amount α, which may specifically include steps S321-S325, Each step is described in detail.

Step S321, which may be referred to as a self-adjusting step, for self-adjusting the transform coefficients of the respective color channels. In a specific embodiment, the maximum transform coefficient c_max is used (if the R color channel has the largest transform coefficient c_max, then at this time, c_r and C_max is equal) decrementing an adjustment step λ, increasing the transform coefficient of a color channel adjacent to the color channel having the largest transform coefficient (such as a G color channel) by an adjustment step λ, according to the transform coefficients of all color channels And the transformation coefficient of the remaining color channel is adjusted for 1 and is expressed by the formula as

C_g=c_g+λ

C_b=1-c_r-c_g

It should be noted that there is another case in this embodiment, when the maximum transform coefficient c_max corresponding to the R color channel is automatically reduced by one adjustment step λ, the transform coefficient of the adjacent B color channel can be automatically adjusted by one. Step size λ, according to the sum of the transform coefficients of all the color channels is 1 and then adjust the transform coefficients of the remaining color channels, which are specifically expressed by the formula

C_b=c_b+λ

C_g=1-c_r-c_b

In another embodiment, one adjustment step λ can be automatically subtracted from the maximum transform coefficient c_max, and the transform coefficients of the remaining two color channels are respectively increased by half adjustment steps, so that all color channels can also be achieved. The requirement that the sum of the transform coefficients is 1, is expressed by a formula

C_g=c_g+λ/2

C_b=c_b+λ/2

It should be understood by those skilled in the art that the self-adjusting method specifically represented by the formula in the above embodiment is the case where the R color channel has the largest transform coefficient c_max, then, by referring to the formulas, the G color channel has the largest transform coefficient and The formula for the B color channel with the largest transform coefficient is not described here. It should also be understood by those skilled in the art that the above embodiments only enumerate the optimal implementation method for the self-increase or self-decrement of the transform coefficients of the respective color channels, in addition to which the self-increase or self-decrement such as λ/4 can be added. For the method of adjusting the value, as long as the sum of the transform coefficients of all the color channels is 1, the specific adjustment value is not limited.

Step S322, calculating and updating the energy of each color channel according to the converted transform coefficients of the respective color channels. In a specific embodiment, the energy of the three color channels R, G, and B are respectively calculated according to step S220, then each The energy of the color channel is updated to En(Y ₁ )', En(Y ₂ )', and En(Y ₃ )', respectively.

Step S323, determining whether the energy of any color channel update is less than the minimum energy in each color channel obtained before the update, and if yes, proceeding to step S324, otherwise proceeding to step S325.

Step S324, updating the energy of all the color channels satisfying the determination condition in step S323 to the minimum energy obtained before the update, for example, before proceeding to step S321, if the R color channel has the minimum energy En(Y ₁ ), when determining En When both (Y ₂ )' and En(Y ₃ )' are smaller than the minimum energy En(Y ₁ ), both En(Y ₂ )' and En(Y ₃ )' are updated to En(Y ₁ ), and the rest are not The energy of the color channel that satisfies the judgment condition remains unchanged.

Step S325, which may be referred to as a second determining step, determining whether the sum of the adjustment step sizes of the maximum transform coefficients is reduced to a preset adjustment amount α. If not, proceeding to step S321; otherwise, outputting the latest adjustment of each color channel. The resulting transform coefficients are re-compared to obtain the maximum transform coefficient c_max.

In a specific embodiment, the adjustment step size λ and the adjustment amount α can be set to 0.02 and 0.2, respectively, and the number of cycles k can be set to record the number of self-adjustment times shown in step S321, and k starts counting from 0, and each step is performed. S321 then k is incremented by 1. When the sum of the adjustment step sizes of the self-decreasing k*0.02 does not reach the adjustment amount of 0.2, step S321 is performed again; when the sum of the adjustment step sizes of the self-decreasing k*0.02 reaches the adjustment amount of 0.2, the latest adjusted transformation is output. The coefficients c_r, c_g, and c_b are compared and the maximum value among the three is compared, and the maximum value is updated to the maximum transform coefficient c_max, and the next iteration operation is performed.

In another embodiment, step S325 can be implemented in another manner of determining. Obtaining the adjustment range [c_max-α, c_max] according to the maximum transform coefficient c_max and the preset adjustment amount α, and determining whether the adjusted maximum transform coefficient c_max exceeds the adjustment range [c_max-α, c_max], and if not, proceeding to step S321 Until the maximum transform coefficient c_max is about to exceed the adjustment range [c_max-α, c_max], otherwise, the newly adjusted transform coefficients c_r, c_g and c_b of the respective color channels are output, and the maximum value among the three is compared, which will be the largest The value is updated to the maximum transform coefficient c_max for the next iteration.

Step S330, which may be referred to as a first determining step, determines whether the preset number of iterations is reached. If not, the process proceeds to step S321, and the iterative operation is performed again. Otherwise, the process proceeds to step S340.

In a specific embodiment, an iteration counter t may be set to record the number of iterations of step S320, t starts counting from 0, and each time step S320 is performed, t is incremented by 1, when t exceeds the number of iterations set by the user (eg 2) When the iterative operation is no longer performed, the process proceeds to step S340.

In step S340, the latest transform coefficients c_r, c_g, and c_b of the respective color channels are output as the optimum coefficients of the corresponding channels.

Step S400, the grayscale image corresponding to the color image is obtained according to the optimal coefficient of each color channel. In an embodiment, the original RGB color is calculated according to the optimal coefficient c_r of the R color channel, the optimal coefficient c_g of the G color channel, and the optimal coefficient c_b of the B color channel. The grayscale image corresponding to the image can be expressed as a formula

y=r*c_r+g*c_g+b*c_b

Wherein, r, g, and b respectively represent the brightness of the R, G, and B color channels on the original RGB color image, and the process of obtaining the brightness of the color channel belongs to the prior art, and is not specifically described herein. It should be noted that when the original color image is directly converted into gradation according to the optimal coefficient of each color channel, it is beneficial to retain most of the image information on the original color image, and the operation process is relatively simple, which can save a lot of calculation time. .

Referring to FIG. 5, in one embodiment, the present application discloses an image gradation device 5, which includes an input unit 51, a color channel acquisition unit 52, an optimal coefficient acquisition unit 53, and a calculation unit. 54, respectively, explained below.

The input unit 51 is configured to input a color image, and the specific process of inputting the color image may refer to step S100, and details are not described herein again.

The color channel acquisition unit 52 is in communication with the input unit 51 for acquiring the energy of each color channel of the color image. For the specific process, refer to step S200, and details are not described herein.

The optimal coefficient acquisition unit 53 is communicatively coupled to the color channel acquisition unit 52 for obtaining the optimal coefficients of the respective color channels according to the energy of the respective color channels. For the specific process, reference may be made to step S300, which is not described herein.

The calculation unit 54 is in communication with the optimal coefficient acquisition unit 53 for obtaining the grayscale image corresponding to the color image according to the optimal coefficient of each color channel. For the specific process, reference may be made to step S400, and details are not described herein.

Further, in another embodiment, the image grading device 5 may further include a display unit 55, which may be communicatively coupled to the computing unit 54 for displaying the grayscale image output by the computing unit 54, even the display unit 55. It is also possible to display the color image processed by the computing unit so that the color image and the grayscale image can be compared, which is convenient for the staff to observe the comparison result. In addition, the display unit 55 can be various types of display devices that can display screens such as televisions, display screens, projectors, and the like.

Those skilled in the art can understand that all or part of the functions of the various methods in the above embodiments may be implemented by hardware or by a computer program. When all or part of the functions in the above embodiments are implemented by a computer program, the program may be stored in a computer readable storage medium, and the storage medium may include: a read only memory, a random access memory, a magnetic disk, an optical disk, a hard disk, etc. The computer executes the program to implement the above functions. For example, the program is stored in the memory of the device, and when the program in the memory is executed by the processor, all or part of the above functions can be realized. In addition, when all or part of the functions in the above embodiment are implemented by a computer program, the program may also be stored in a storage medium such as a server, another computer, a magnetic disk, an optical disk, a flash disk or a mobile hard disk, and may be saved by downloading or copying. The system is updated in the memory of the local device, or the system of the local device is updated. When the program in the memory is executed by the processor, all or part of the functions in the above embodiments may be implemented.

The invention has been described above with reference to specific examples, which are merely intended to aid the understanding of the invention and are not intended to limit the invention. For the person skilled in the art to which the invention pertains, several simple derivations, variations or substitutions can be made in accordance with the inventive concept.

Claims (11)

1. An image grayscale method, comprising the steps of:

   Enter a color image;

   Obtaining energy of each color channel of the color image separately;

   Obtaining an optimum coefficient of each color channel according to the energy of each color channel;

   Obtaining a grayscale image corresponding to the color image according to an optimal coefficient of each color channel.
2. The image gradation method according to claim 1, wherein the respectively acquiring energy of each color channel of the color image comprises:

   Performing downsampling processing on the color image to obtain a scaled image;

   Obtaining an image color distribution of the scaled image, and a probability histogram of each color channel;

   The energy corresponding to each color channel is obtained according to the image color distribution and the probability histogram of each color channel.
3. The method of image grading according to claim 2, wherein the downsampling the color image to obtain a zoomed image comprises:

   Obtaining a height pixel h and a width pixel w of the color image;

   When h is smaller than the first value and w is smaller than the second value, the color image is taken as the zoomed image;

   When h is greater than or equal to the first value and less than or equal to the third value, or w is greater than or equal to the second value and less than or equal to the fourth value, w and h are each reduced to 1/2 of the original, and the scaled image is used as the Scale the image;

   When h is greater than the third value or w is greater than the fourth value, w and h are each reduced to the original 1/4, and the scaled image is taken as the scaled image.
4. The image gradation method according to claim 2, wherein the obtaining an image color distribution according to the zoomed image comprises:

   Performing cluster analysis on the scaled image to represent similar colors on the scaled image with the same color;

   And quantizing the scaled image into a color class diagram according to a result of cluster analysis;

   The image color distribution is obtained from the color class diagram.
5. The image gradation method according to claim 1, wherein the obtaining the optimal coefficients of the respective color channels according to the energy of the respective color channels comprises:

   Energy and coefficient acquisition step: comparing energy of each color channel to obtain minimum energy; respectively obtaining transform coefficients of each color channel, and using a transform coefficient of a color channel corresponding to the minimum energy as a maximum transform coefficient;

   An iterative step: updating the transform coefficients of the respective color channels according to the maximum transform coefficient and the preset adjustment step size and the adjustment amount;

   The first determining step: determining whether the preset number of iterations is reached, and if not, repeating the iterative step, and vice versa, outputting the latest transform coefficient of each color channel as the optimal coefficient of the corresponding channel.
6. The image grayscale method according to claim 5, wherein the iterative step comprises:

   Self-adjusting step: the maximum transform coefficient is automatically reduced by one adjustment step, and the transform coefficient of one color channel adjacent to the color channel having the largest transform coefficient is incremented by one adjustment step, and the sum of the transform coefficients according to all color channels is 1 Then, the transform coefficients of the remaining color channels are adjusted; the energy of each color channel is calculated and updated according to the adjusted transform coefficients of the respective color channels, and the energy after updating any color channel is determined to be smaller than the minimum of each color channel obtained before the update. When energy is used, the energy of the color channel is updated to the minimum energy obtained before the update;

   a second determining step: determining whether the sum of the adjustment step lengths of the maximum transform coefficients is up to a preset adjustment amount, and if not, repeating the self-adjusting step, and vice versa, outputting the latest adjusted transform coefficients of the respective color channels And compare to get the maximum transform coefficient.
7. The image grayscale method according to claim 6, wherein the second determining step comprises:

   Obtaining an adjustment range according to the maximum transform coefficient and the preset adjustment amount;

   Determining whether the adjusted maximum transform coefficient exceeds the adjustment range;

   If not, the self-adjusting step is repeated until the maximum transform coefficient is about to exceed the adjustment range. Otherwise, the newly adjusted transform coefficients of the respective color channels are outputted, and the maximum transform coefficients are compared.
8. The image gradation method according to claim 1, wherein the obtaining the grayscale image corresponding to the color image according to the optimal coefficient of each color channel comprises:

   The grayscale image corresponding to the color image is calculated according to the optimal coefficient of each color channel and the brightness corresponding to each color channel.
9. An image graying device, comprising:

   An input unit for inputting a color image;

   a color channel acquiring unit, configured to respectively acquire energy of each color channel of the color image;

   An optimal coefficient acquisition unit, configured to obtain an optimal coefficient of each color channel according to the energy of each color channel;

   And a calculating unit, configured to obtain a grayscale image corresponding to the color image according to an optimal coefficient of each color channel.
10. The image gradation device according to claim 9, further comprising a display unit for displaying said grayscale image and/or said color image.
11. A computer readable storage medium, comprising a program executable by a processor to implement the method of any of claims 1-8.

Images