CN116402729A - An image enhancement method and system based on double histogram equalization - Google Patents

Abstract

The invention discloses an image enhancement method and system based on double histogram equalization, and relates to the technical field of image processing. Firstly, grayscale processing is performed on the image to be processed, and the grayscale histogram is obtained by statistics; the optimal segmentation is determined by an improved Otsu algorithm points to get the target area A and background area B; determine the segmentation threshold T _A of the target area A and the segmentation threshold T B of the background area _B ; independently equalize the gray sub-histogram; use the segmentation threshold T _A and T _B to The above equalized gray histogram is over-enhanced and suppressed, and local gray correction is performed to obtain the enhanced image. The invention can enhance the image contrast and detail information while protecting the average brightness, avoid noise amplification as much as possible, has high robustness, strong stability, wide applicability, and significantly improves image quality.

Description

An image enhancement method and system based on double histogram equalization

technical field

The present invention relates to the technical field of image processing, and more specifically relates to an image enhancement method and system based on double histogram equalization.

Background technique

my country's industrial manufacturing industry is transforming towards electrification, networking, and intelligence. Automatic measurement based on machine vision can not only shorten the measurement time, but also save labor costs. It is an important way to improve efficiency while ensuring quality. But this also places higher demands on the accuracy, speed and automation level of the measurement system. It is necessary to continuously improve and optimize the detection technology to adapt to the rapid development of science and technology, and explore faster, more accurate, more efficient and cost-effective detection methods.

In the industrial field, the measurement of the target can be quickly realized through the machine vision system. However, in actual work, due to environmental problems (rain, fog, smoke, etc.), imaging equipment problems, and lighting problems, the image quality obtained is low and the contrast is low. Low, not conducive to subsequent image processing.

Nanjing University of Science and Technology invented a histogram equalization method based on block statistics (CN 110223241A). First, the input image is divided into blocks on average, and divided into M*N sub-images; then the histogram statistics are performed on each sub-image, and the judgment conditions are set during the statistics, and different histogram statistics are performed on the uniform scene and the rich scene. Obtain the histogram mapping function; then perform bilinear interpolation on the pixels of the original image according to the position and the value of the histogram mapping function of the adjacent block to obtain the final mapping value. The specific process is shown in Figure 1. However, this method performs global histogram equalization on the image, which is prone to brightness shift. After image enhancement, it cannot ensure the enhancement of image details while improving the contrast.

Huaqiao University invented a global histogram equalization method with adaptive threshold (CN 109801246A). Statistically and scale the histogram data so that the mean value of the histogram is 1. After preprocessing, calculate the adaptive optimal truncation threshold, perform truncation and post-processing on the histogram according to the optimal truncation threshold, and calculate the mapping table; The table lookup operation is performed to finally obtain a gray scale enhanced image. The specific process is shown in Figure 2. This method averagely divides the input image into blocks. Although it can effectively avoid the loss of image details, it is prone to noise and "block effect", which leads to local distortion of the processed image, and the processing effect is not ideal.

In the patented FPGA-based low-caching improved histogram equalization method and system (CN 110148101B), the time-division multiplexing control signal is generated by the time-division multiplexing module, and the buffer control module generates three histograms. Statistical branch buffer timing control Signal, to solve the problems of noise amplification, image distortion and loss of details in traditional image enhancement methods, but this method is highly targeted and only applicable to visible light and infrared image enhancement processing, and its real-time performance is poor.

Therefore, based on the technical defects in the image enhancement processing in the prior art, how to provide an image enhancement method that can enhance the image contrast and detail information while protecting the average brightness, and significantly improve the image quality is a technology in the art. Problems that people need to solve urgently.

Contents of the invention

In view of this, the present invention provides an image enhancement method and system based on double histogram equalization.

In order to achieve the above object, the present invention provides the following technical solutions:

An image enhancement method based on double histogram equalization, comprising the following steps:

Step 1, obtain the image to be processed;

Step 2, performing grayscale processing on the image to be processed to obtain an original grayscale image F, and performing grayscale statistics on the original grayscale image to obtain a grayscale histogram;

Step 3. Determine the optimal segmentation point K _out of the original grayscale image F, and divide the original grayscale image F into the target area A and the background area B according to the gray level based on the optimal segmentation point K _out ;

Step 4. Determine the segmentation threshold T _A of the grayscale sub-histogram of the target area A and the segmentation threshold T _B of the grayscale sub-histogram of the background area B based on the optimal segmentation point K _out ;

Step 5. Independently equalize the gray level sub-histogram of the target area A and the gray level sub-histogram of the background area B to obtain the equalized gray level histogram P _S ;

Step 6. Based on the segmentation threshold _TA and the segmentation threshold _TB , perform over-enhancement suppression on the equalized gray histogram _PS to obtain an image _PT ;

Step 7. Perform local grayscale correction on the image _PT to obtain an enhanced image.

Optionally, in the step 3, an improved Otsu algorithm is used to determine the optimal segmentation point K _out of the original grayscale image F, and the specific method is as follows:

K _out = arg _t max G(t);

表示目标区域A平均方差；/>

表示背景区域B平均方差。Among them, q _A (t) indicates the proportion of the target area A when the threshold t is the segmentation point; q _B (t) indicates the proportion of the background area B when the threshold t is the segmentation point; μ _A (t) indicates the threshold t When is the segmentation point, the probability of the target area A; μ _B (t) represents the probability of the background area B when the threshold t is the segmentation point;

Indicates the average variance of the target area A; />

Indicates the mean variance of the background region B.

Optionally, in the step 3, the target area A is composed of pixels whose gray values are in the interval [MIN, K _out ], and the background area B is composed of pixels whose gray values are in the interval [K _out +1, MAX] , MIN represents the minimum pixel value in the original grayscale image F, and MAX represents the maximum pixel value in the original grayscale image F.

Optionally, in step 4, the method for determining the segmentation threshold _TA and the segmentation threshold _TB is:

Among them, _TMA represents the peak value of the grayscale sub-histogram of the target area A, _TMB represents the peak value of the grayscale sub-histogram of the background area B, MIN represents the minimum pixel value in the original grayscale image F, and MAX represents the original grayscale The maximum pixel value in image F.

Optionally, in step 5, the method for independently equalizing the grayscale sub-histogram of the target area A and the grayscale sub-histogram of the background area B is:

Among them, n _i represents the total number of pixels of gray level i in the original gray scale image F, n represents the total number of all pixels in the original gray scale image F, P _S (i) represents the gray level histogram after equalization when the gray level is i In the figure, NA represents the total number of gray levels in the target area A, NB represents the total number of gray levels in the background area B, MIN represents the minimum pixel value in the original grayscale image F, and MAX represents the maximum pixel value in the original grayscale image F.

Optionally, in the step 6, based on the segmentation threshold T _A and the segmentation threshold T _B , the method of suppressing the over-enhancement of the equalized grayscale histogram _PS is as follows:

Among them, P _T (i) represents the gray histogram after clipping when the gray level is i.

Optionally, in step 7, the method for performing local grayscale correction on the image _PT is:

Correct the gray value of the center pixel of the image _PT , specifically,

表示步骤5独立均衡后的灰度图像中以(i,j)为中心的5×5窗口内像素的灰度平均值，D_in表示原始灰度图像F的梯度矩阵，D_HE表示步骤5独立均衡后的灰度图像的梯度矩阵。Among them, x _out (i, j) represents the gray value of the central pixel of the corrected image _PT , x _in (i, j) represents the gray value of the central pixel of the original gray image F, and x _HE (i, j) represents the gray value of the image The gray value of the central pixel of P _T , x _main (i, j) is the average gray value of the pixels in the 5×5 window centered on (i, j) in the original gray image F,

Indicates the gray average value of the pixels in the 5×5 window centered on (i, j) in the gray image after step 5 independent equalization, D _in represents the gradient matrix of the original gray image F, D _HE represents the independent The gradient matrix of the equalized grayscale image.

Optionally, use the sobel operator to perform gradient matrix convolution on the original grayscale image F to obtain the gradient matrix D _in :

Use the sobel operator to perform gradient matrix convolution on the independently equalized grayscale image to obtain the gradient matrix D _HE :

Among them, D _0° , D _180° , D _45° , and D _135° represent the sobel operators in the directions of 0°, 180°, 45°, and 135°, respectively.

Optionally, before determining the segmentation threshold T _A and T _B in step 4 and performing independent equalization in step 5, it is also necessary to perform one-dimensional Median filtering, the specific method is:

Obtain the sets F A and F _B of non-zero units in the gray sub-histogram of the target area A and the gray sub-histogram of the background area _B respectively:

Among them, i represents the pixel gray level;

A 3×3 one-dimensional median filter is performed on the sets F _A and F _B .

The invention also discloses an image enhancement system based on double histogram equalization, including:

An image acquisition module, configured to acquire images to be processed;

A grayscale processing module, configured to perform grayscale processing on the image to be processed to obtain an original grayscale image F, and perform grayscale statistics on the original grayscale image to obtain a grayscale histogram;

The image division module is used to determine the optimal segmentation point K _out of the original grayscale image F, and divide the original grayscale image F into the target area A and the background area B according to the gray level based on the optimal segmentation point K _out ;

A segmentation threshold determination module, configured to determine the segmentation threshold T A of the grayscale sub-histogram of the target area A based on the optimal segmentation point K _out , and the segmentation threshold T _B of the grayscale sub-histogram of the background area B _;

An independent equalization processing module, which is used to independently equalize the gray level sub-histogram of the target area A and the gray level sub-histogram of the background area B to obtain the equalized gray level histogram P _S ;

An over-enhancement suppression module, configured to perform over-enhancement suppression on the equalized grayscale histogram _PS based on the segmentation threshold _TA and the segmentation threshold _TB to obtain an image _PT ;

The local grayscale correction module is configured to perform local grayscale correction on the image _PT to obtain an enhanced image.

It can be seen from the above technical solutions that the present invention provides an image enhancement method and system based on double histogram equalization, which has the following beneficial effects compared with the prior art:

(1) The global histogram equalization algorithm will have a brightness shift during the image equalization process, resulting in the loss of image detail information. Therefore, the present invention uses the improved Otsu algorithm to divide the image into two sub-histograms of the target and the background and perform independent equalization to protect the average brightness of the input image and avoid the problem of image brightness shift after enhancement.

(2) Compared with the traditional Otsu method, the image segmented by the improved Otsu algorithm of the present invention can show more image details and achieve a better segmentation effect.

(3) The traditional histogram equalization algorithm will over-enhance the image due to over-stretching, resulting in an unnatural enhanced image. Therefore, the present invention adaptively obtains the threshold value to suppress the gray level to avoid excessive enhancement of the image, and the improved algorithm has a protective effect on the edge details of the image, and the image quality is significantly improved.

(4) Using the local gray-scale correction algorithm for gray-scale correction, effectively avoiding the problem of gray-scale merging, protecting image detail information, image contrast and edge detail information have been reasonably enhanced, image details are clearer, and have better The visual effect can effectively improve the accuracy of machine vision measurement in low-light environment.

In summary, the present invention can enhance the image contrast and detail information while protecting the average brightness, and at the same time avoid the amplification of noise as much as possible, with high robustness, strong stability, wide applicability, and significantly improved image quality .

The invention is especially suitable for low-illuminance image enhancement in low-illuminance environments such as factories and mines, so as to measure more accurate industrial data, promote the implementation of enterprise engineering projects, and improve production efficiency.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the 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 It is an embodiment of the present invention, and those skilled in the art can also obtain other drawings according to the provided drawings without creative work.

Fig. 1 is a method schematic diagram of a histogram equalization method (CN 110223241 A) based on block statistics of a disclosed patent;

Fig. 2 is a method flow diagram of a global histogram equalization method (CN 109801246A) of a published patent with adaptive threshold;

Fig. 3 is a flow chart of an image enhancement method based on double histogram equalization provided by the present invention;

Fig. 4 is a schematic diagram of an image enhancement system module based on double histogram equalization provided by the present invention;

Figure 5(a) is the original image of scene 1;

Figure 5(b) is the rendering of the original image of scene 1 processed by the traditional Otsu algorithm;

Figure 5(c) is the rendering of the original image of scene 1 processed by the improved Otsu algorithm;

Figure 5(d) is the original image of scene 2;

Fig. 5 (e) is the effect diagram after the original image of scene 2 is processed by the traditional Otsu algorithm;

Fig. 5(f) is the effect diagram after the original image of scene 2 is processed by the improved Otsu algorithm;

Fig. 6 (a) is the original gray scale image in the embodiment of the present invention;

Fig. 6 (b) is the grayscale histogram of original grayscale image in the embodiment of the present invention;

Fig. 7 (a) is the image after image enhancement in the embodiment of the present invention;

Fig. 7(b) is a histogram of image grayscale after image enhancement in the embodiment of the present invention.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

The embodiment of the present invention discloses an image enhancement method based on dual histogram equalization, which can be applied to image enhancement of images containing detection targets in low-illuminance environments in the industrial field, so as to realize industrial measurement in low-illuminance environments and improve industrial production. efficiency. Referring to Figure 3, it specifically includes the following steps:

Step 1. Obtain the image to be processed.

The image data of the target to be detected in a low-light environment is obtained by shooting with an industrial camera.

Step 2. Perform grayscale processing on the image to be processed to obtain an original grayscale image F, and perform grayscale statistics on the original grayscale image to obtain a grayscale histogram.

Step 3. Use the improved Otsu algorithm to determine the optimal segmentation point K _out of the original grayscale image F, and divide the original grayscale image F into the target area A and the background area B according to the gray level based on the optimal segmentation point K _out .

Among them, the specific process of using the improved Otsu algorithm to determine the best segmentation point _Kout of the original grayscale image F is:

Set the threshold t as the segmentation point, and divide the image into the target area A and the background area B according to the gray level. The target area A is composed of pixels with gray values in the interval [MIN,t], and the background area B is composed of pixels with gray values in [ t+1,MAX] pixels in the interval. Then the ratio q _A (t) and q _B (t) of class A to class B is:

Among them, L is the number of gray levels, and P(i) represents the probability when the gray value is i. According to the probability of class A and class B, it can be expressed by μ _A (t) and μ _B (t):

Then find the average gray value μ of the input image:

μ= _qA (t) _μA (t)+ _qB (t) _μB (t);

定义为：between-class variance

defined as:

To measure the quality of pixel cohesion, assume a distance metric, namely:

d ² (t) = (μ _A (t) - μ _B (t)) ² ;

Introduce the average variance of the target area and the background area

Get the new threshold calculation formula:

When G(t) takes the maximum value, the corresponding t value is the best segmentation point. Therefore, the formula for calculating the optimal segmentation point K _out is as follows:

K _out = arg _t max G(t).

Based on the optimal segmentation point K _out , the original grayscale image F is divided into a target area A and a background area B according to the gray level, the target area A is composed of pixels whose gray values are in the interval [MIN, K _out ], and the background area B is composed of pixels whose grayscale value is in the interval [K _out +1, MAX], MIN represents the minimum pixel value in the original grayscale image F, and MAX represents the maximum pixel value in the original grayscale image F.

Referring to Fig. 5(a)-Fig. 5(c) is a comparison diagram of the effect after scene 1 adopts traditional Otsu algorithm and the improved Otsu algorithm of the present invention, and Fig. 5(d)-Fig. 5(f) is scene 2 adopting traditional Otsu The comparison chart of the algorithm and the improved Otsu algorithm of the present invention after processing shows that the improved Otsu algorithm proposed by the present invention can display more image details and achieve a better segmentation effect.

In the specific application process, it is also necessary to perform one-dimensional median filtering on the gray level sub-histogram of the target area A and the gray level sub-histogram of the background area B to facilitate the next step of processing. The specific method is as follows:

Obtain the sets F A and F _B of non-zero units in the gray sub-histogram of the target area A and the gray sub-histogram of the background area _B respectively:

Among them, i represents the pixel gray level;

A 3×3 one-dimensional median filter is performed on the sets F _A and F _B .

Step 4. Determine the segmentation threshold T _A of the grayscale sub-histogram of the target area A and the segmentation threshold T _B of the grayscale sub-histogram of the background area B based on the optimal segmentation point K _out :

Among them, _TMA represents the peak value of the grayscale sub-histogram of the target area A, _TMB represents the peak value of the grayscale sub-histogram of the background area B, MIN represents the minimum pixel value in the original grayscale image F, and MAX represents the original grayscale The maximum pixel value in image F.

Step 5. Independently equalize the gray level sub-histogram of the target area A and the gray level sub-histogram of the background area B, and count the equalized image gray level to obtain the equalized gray level histogram P _S . The equalization formula is :

Among them, n _i represents the total number of pixels of gray level i in the original gray scale image F, n represents the total number of all pixels in the original gray scale image F, P _S (i) represents the gray level histogram after equalization when the gray level is i In the figure, NA represents the total number of gray levels in the target area A, NB represents the total number of gray levels in the background area B, MIN represents the minimum pixel value in the original grayscale image F, and MAX represents the maximum pixel value in the original grayscale image F.

Step 6. Based on the segmentation threshold T _A and the segmentation threshold T _B , perform over-enhancement suppression on the equalized grayscale histogram _PS , that is, in the target area A, values greater than the segmentation threshold T _A become is T _A , in the background area B, the value greater than the segmentation threshold T _B becomes T _B , and the rest of the values remain unchanged, and the image P _T is obtained:

Among them, P _T (i) represents the gray histogram after clipping when the gray level is i.

Step 7. Perform local gray scale correction on the image _PT to obtain an enhanced image:

表示步骤5独立均衡后的灰度图像中以(i,j)为中心的5×5窗口内像素的灰度平均值，D_in表示原始灰度图像F的梯度矩阵，D_HE表示步骤5独立均衡后的灰度图像的梯度矩阵。Among them, x _out (i, j) represents the gray value of the central pixel of the corrected image _PT , x _in (i, j) represents the gray value of the central pixel of the original gray image F, and x _HE (i, j) represents the gray value of the image The gray value of the central pixel of P _T , x _main (i, j) is the average gray value of the pixels in the 5×5 window centered on (i, j) in the original gray image F,

Indicates the gray average value of the pixels in the 5×5 window centered on (i, j) in the gray image after step 5 independent equalization, D _in represents the gradient matrix of the original gray image F, D _HE represents the independent The gradient matrix of the equalized grayscale image.

Use the sobel operator to perform gradient matrix convolution on the original grayscale image F to obtain the gradient matrix D _in :

Use the sobel operator to perform gradient matrix convolution on the independently equalized grayscale image to obtain the gradient matrix D _HE :

Among them, D _0° , D _180° , D _45° , and D _135° represent the sobel operators in the directions of 0°, 180°, 45°, and 135°, respectively.

Another embodiment of the present invention also discloses an image enhancement system based on double histogram equalization, see Figure 4, including:

An image acquisition module, configured to acquire images to be processed;

A grayscale processing module, configured to perform grayscale processing on the image to be processed to obtain an original grayscale image F, and perform grayscale statistics on the original grayscale image to obtain a grayscale histogram;

The image division module is used to determine the optimal segmentation point K _out of the original grayscale image F, and divide the original grayscale image F into the target area A and the background area B according to the gray level based on the optimal segmentation point K _out ;

A segmentation threshold determination module, configured to determine the segmentation threshold T A of the grayscale sub-histogram of the target area A based on the optimal segmentation point K _out , and the segmentation threshold T _B of the grayscale sub-histogram of the background area B _;

An independent equalization processing module, which is used to independently equalize the gray level sub-histogram of the target area A and the gray level sub-histogram of the background area B to obtain the equalized gray level histogram P _S ;

An over-enhancement suppression module, configured to perform over-enhancement suppression on the equalized grayscale histogram _PS based on the segmentation threshold _TA and the segmentation threshold _TB to obtain an image _PT ;

The local grayscale correction module is configured to perform local grayscale correction on the image _PT to obtain an enhanced image.

In the industrial field, the measurement of the target can be quickly realized through the machine vision system. However, in actual work, due to environmental problems (rain, fog, smoke, etc.), imaging equipment problems, and lighting problems, the image quality obtained is low and the contrast is low. Low, it is not conducive to subsequent image processing, so the image contrast can be improved by the image enhancement technology of the present invention, so as to finally obtain more accurate industrial data and improve the robustness of the automatic detection system.

Similarly, in addition to the above-mentioned industrial fields, the technical solution of the present invention can also be applied to other scenarios requiring image enhancement. For example, the technical solution of the present invention can also be applied to visible light and infrared image enhancement processing, and infrared imaging technology is widely used in the fields of national defense, security, non-destructive testing, and poisonous gas detection. For example, in the automatic visual inspection system of faulty components, faulty components usually generate too much heat. In assembly, infrared images are generated from the distribution of heat energy. By enhancing the processing of infrared images, faulty components can be accurately identified in assembly. Improve image quality, enhance image details, and improve the robustness of the detection system.

Further, the present invention can also be applied to lung disease identification, heart disease identification, digital mammogram, etc., to realize image enhancement in biomedicine. Since the original image obtained by the imaging device is affected by various factors such as the hardware performance of the device itself and the acquisition conditions, the medical image obtained directly from the medical instrument may have image quality degradation, such as low contrast, unclear image, etc. Through the present invention, the contrast of medical images can be improved, thereby assisting doctors in automatic diagnosis and improving work efficiency.

Referring to Fig. 6 (a), Fig. 6 (b) is the original grayscale image of a low-illuminance image and its grayscale histogram, Fig. 7 (a) and Fig. 7 (b) adopt the present invention for described low-illuminance image The image and its grayscale histogram after image enhancement by the scheme, it can be seen that the present invention can enhance the image contrast and detail information while protecting the average brightness, and at the same time avoid the amplification of noise as much as possible and improve the image quality.

Each embodiment in this specification is described in a progressive manner, each embodiment focuses on the difference from other embodiments, and the same and similar parts of each embodiment can be referred to each other. As for the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and for relevant details, please refer to the description of the method part.

The above description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (10)

1. an image enhancement method based on double histogram equalization, is characterized in that, comprises the following steps: Step 1, obtain the image to be processed; Step 2, performing grayscale processing on the image to be processed to obtain an original grayscale image F, and performing grayscale statistics on the original grayscale image to obtain a grayscale histogram; Step 3. Determine the optimal segmentation point K _out of the original grayscale image F, and divide the original grayscale image F into the target area A and the background area B according to the gray level based on the optimal segmentation point K _out ; Step 4. Determine the segmentation threshold T _A of the grayscale sub-histogram of the target area A and the segmentation threshold T _B of the grayscale sub-histogram of the background area B based on the optimal segmentation point K _out ; Step 5. Independently equalize the gray level sub-histogram of the target area A and the gray level sub-histogram of the background area B to obtain the equalized gray level histogram P _S ; Step 6. Based on the segmentation threshold _TA and the segmentation threshold _TB , perform over-enhancement suppression on the equalized gray histogram _PS to obtain an image _PT ; Step 7. Perform local grayscale correction on the image _PT to obtain an enhanced image. 2. a kind of image enhancement method based on double histogram equalization according to claim 1, is characterized in that, in described step 3, adopts improved Otsu algorithm to determine the optimal segmentation point K _out of original grayscale image F, The specific method is as follows: K _out = arg _t maxG(t);

Among them, q _A (t) indicates the proportion of the target area A when the threshold t is the segmentation point; q _B (t) indicates the proportion of the background area B when the threshold t is the segmentation point; μ _A (t) indicates the threshold t When is the segmentation point, the probability of the target area A; μ _B (t) represents the probability of the background area B when the threshold t is the segmentation point;

Indicates the average variance of the target area A; />

Indicates the mean variance of the background region B. 3. a kind of image enhancement method based on double histogram equalization according to claim 1, is characterized in that, in described step 3, target area A is made up of the pixel of grayscale value in [MIN, K _out ] interval , the background area B is composed of pixels whose grayscale values are in the interval [K _out +1, MAX], MIN represents the minimum pixel value in the original grayscale image F, and MAX represents the maximum pixel value in the original grayscale image F. 4. a kind of image enhancement method based on double histogram equalization according to claim 1, is characterized in that, in described step 4, the method for determining segmentation threshold _TA and segmentation threshold _TB is:

Among them, _TMA represents the peak value of the grayscale sub-histogram of the target area A, _TMB represents the peak value of the grayscale sub-histogram of the background area B, MIN represents the minimum pixel value in the original grayscale image F, and MAX represents the original grayscale The maximum pixel value in image F. 5. A kind of image enhancement method based on double histogram equalization according to claim 1, characterized in that, in the step 5, the grayscale sub-histogram of the target area A and the grayscale sub-histogram of the background area B The method for independent equalization of the histogram is:

Among them, n _i represents the total number of pixels of gray level i in the original gray scale image F, n represents the total number of all pixels in the original gray scale image F, P _S (i) represents the gray level histogram after equalization when the gray level is i In the figure, NA represents the total number of gray levels in the target area A, NB represents the total number of gray levels in the background area B, MIN represents the minimum pixel value in the original grayscale image F, and MAX represents the maximum pixel value in the original grayscale image F. 6. A kind of image enhancement method based on dual histogram equalization according to claim 5, characterized in that, in the step 6, based on the segmentation threshold _TA and the segmentation threshold _TB , the The method of over-enhanced suppression of the equalized gray histogram _PS is:

Among them, P _T (i) represents the gray histogram after clipping when the gray level is i. 7. a kind of image enhancement method based on double histogram equalization according to claim 1, is characterized in that, in described step 7, the method that described image _PT is carried out local grayscale correction is: Correct the gray value of the center pixel of the image _PT , specifically,

Among them, x _out (i, j) represents the gray value of the central pixel of the corrected image _PT , x _in (i, j) represents the gray value of the central pixel of the original gray image F, and x _HE (i, j) represents the gray value of the image The gray value of the central pixel of P _T , x _main (i, j) is the average gray value of the pixels in the 5×5 window centered on (i, j) in the original gray image F,

Indicates the gray average value of the pixels in the 5×5 window centered on (i, j) in the gray image after step 5 independent equalization, D _in represents the gradient matrix of the original gray image F, D _HE represents the independent The gradient matrix of the equalized grayscale image. 8. a kind of image enhancement method based on double histogram equalization according to claim 7, is characterized in that, uses sobel operator to carry out gradient matrix convolution to original grayscale image F, obtains gradient matrix D _in :

Use the sobel operator to perform gradient matrix convolution on the independently equalized grayscale image to obtain the gradient matrix D _HE :

Among them, D _0° , D _180° , D _45° , and D _135° represent the sobel operators in the directions of 0°, 180°, 45°, and 135°, respectively. 9. A kind of image enhancement method based on double histogram equalization according to claim 1, it is characterized in that, before step 4 determines segmentation threshold _TA , segmentation threshold _TB and step 5 carries out independent equalization, also need to The gray level sub-histogram of the target area A and the gray level sub-histogram of the background area B are subjected to one-dimensional median filtering, and the specific method is as follows: Obtain the sets F A and F _B of non-zero units in the gray sub-histogram of the target area A and the gray sub-histogram of the background area _B respectively:

Among them, i represents the pixel gray level; A 3×3 one-dimensional median filter is performed on the sets F _A and F _B . 10. An image enhancement system based on double histogram equalization, characterized in that, comprising: An image acquisition module, configured to acquire images to be processed; A grayscale processing module, configured to perform grayscale processing on the image to be processed to obtain an original grayscale image F, and perform grayscale statistics on the original grayscale image to obtain a grayscale histogram; The image division module is used to determine the optimal segmentation point K _out of the original grayscale image F, and divide the original grayscale image F into the target area A and the background area B according to the gray level based on the optimal segmentation point K _out ; A segmentation threshold determination module, configured to determine the segmentation threshold T A of the grayscale sub-histogram of the target area A based on the optimal segmentation point K _out , and the segmentation threshold T _B of the grayscale sub-histogram of the background area B _; An independent equalization processing module, which is used to independently equalize the gray level sub-histogram of the target area A and the gray level sub-histogram of the background area B to obtain the equalized gray level histogram P _S ; An over-enhancement suppression module, configured to perform over-enhancement suppression on the equalized grayscale histogram _PS based on the segmentation threshold _TA and the segmentation threshold _TB to obtain an image _PT ; The local grayscale correction module is configured to perform local grayscale correction on the image _PT to obtain an enhanced image.

Images