CN117408890A

CN117408890A - Video image transmission quality enhancement method and system

Info

Publication number: CN117408890A
Application number: CN202311714491.1A
Authority: CN
Inventors: 查乾
Original assignee: Wuhan Zeta Cloud Technology Co ltd
Current assignee: Wuhan Zeta Cloud Technology Co ltd
Priority date: 2023-12-14
Filing date: 2023-12-14
Publication date: 2024-01-16
Anticipated expiration: 2043-12-14
Also published as: CN117408890B

Abstract

The invention relates to the technical field of image enhancement, in particular to a video image transmission quality enhancement method and a video image transmission quality enhancement system. Obtaining image frames, obtaining gray gradient directions of edge lines in each image frame and pixel points, obtaining illumination influence values according to gray differences among the pixel points in a preset first range, screening pixel points to be denoised according to gray changes of the pixel points in adjacent frames and positions of the edge lines, selecting the pixel points similar to the pixel points to be denoised as reference points, obtaining noise representation values of the reference points, determining gray value weights of the reference points according to the noise representation values of the reference points, gray value change conditions in the adjacent image frames and position relations of the reference points and the pixel points to be denoised, denoising the pixel points to be denoised according to gray values of the reference points in the image frames and gray value weights corresponding to the reference points, and obtaining enhanced images and transmitting the enhanced images. The invention improves the efficiency of image enhancement while ensuring the image transmission quality.

Description

Video image transmission quality enhancement method and system

Technical Field

The invention relates to the technical field of image enhancement, in particular to a video image transmission quality enhancement method and a video image transmission quality enhancement system.

Background

The high-quality video image transmission can provide clearer and truer visual experience for users, and the enhancement of the video image quality plays an important role in improving the performance and user experience of various applications, so that the continuous innovation and development of related technologies are promoted.

Image denoising is a common method for enhancing image quality, and in the prior art, non-local mean filtering is generally adopted to denoise video images so as to improve the image quality, but because the number of video images is large and areas with different change degrees exist, if global filtering is carried out on the video images, the problems of poor image enhancement effect and low efficiency are caused.

Disclosure of Invention

In order to solve the technical problems of poor image enhancement effect and low efficiency caused by the fact that the number of video images is large and areas with different change degrees exist when the non-local mean filtering is adopted for enhancing the image quality in the prior art, the invention aims to provide a video image transmission quality enhancement method and a video image transmission quality enhancement system, and the adopted technical scheme is as follows:

acquiring image frames based on a preset time interval, and acquiring an edge line in each image frame and a gray gradient direction of each pixel point;

Optionally selecting one of the image frames as a target frame; in a target frame, in a preset first range with each pixel point as a center, obtaining an illumination influence value of the center pixel point according to gray level differences among the pixel points on a straight line where gray level gradient directions of all the pixel points are located; screening suspected denoising pixel points in the target frame according to the gray level change of the pixel points at the same position in the target frame and the adjacent image frames; obtaining pixel points to be denoised according to a preset time interval, the position distribution of each suspected denoising pixel point and the edge line and the illumination influence value of each suspected denoising pixel point;

screening reference points according to the difference of illumination influence values of the pixel points and the gray level difference of the pixel points at the same position in the target frame and the adjacent image frames in a preset second range taking each pixel point to be denoised as a center; in a preset neighborhood taking each reference point as a center, obtaining a noise representation value of the reference point according to the gray level difference between each reference point and all neighborhood pixel points; obtaining gray value weight of each reference point according to noise representation values of the reference points, position distribution of the reference points and corresponding pixel points to be denoised and gray value change conditions of the pixel points of the reference points at the same positions in the target frame and the adjacent image frames;

And updating the gray value of each pixel point to be denoised according to the gray value weights of all the reference points corresponding to each pixel point to be denoised and the gray values of the pixel points of the reference points at the same position in the target frame and the adjacent image frames, so as to obtain and transmit the enhanced image.

Further, in the preset first range with each pixel point as a center, obtaining the illumination influence value of the center pixel point according to the gray scale difference between the pixel points on the straight line where the gray scale gradient directions of all the pixel points are located, including:

in a preset first range corresponding to each pixel point, taking a preset number of pixel points on the left side and the right side of each pixel point as target points on a straight line where the gray gradient direction of each pixel point is located; calculating accumulated values of gray value differences between every two target points of each pixel point to obtain illumination influence factors of each pixel point;

and taking other pixel points except the central pixel point in each preset first range as comparison points of the central pixel point, and carrying out negative correlation mapping and normalization on the value obtained by accumulating the differences of the illumination influence factors between all the comparison points and the corresponding central pixel points to obtain the illumination influence value of the central pixel point.

Further, the obtaining the pixel to be denoised according to the preset time interval, the position distribution of each suspected denoised pixel and the edge line, and the illumination influence value of each suspected denoised pixel includes:

taking the shortest distance between each suspected denoising pixel point and the edge line closest to the suspected denoising pixel point as a distance parameter of each suspected denoising pixel point;

obtaining denoising possibility of each suspected denoising pixel point according to the distance parameter, the illumination influence value and the preset time interval of each suspected denoising pixel point, wherein the denoising possibility is positively correlated with the preset time interval, and the illumination influence value and the distance parameter are negatively correlated with the denoising possibility;

and taking the suspected denoising pixel point with the denoising possibility larger than the preset denoising threshold value as the pixel point to be denoised.

Further, in the second preset range centered on each pixel to be denoised, screening the reference point according to the difference of the illumination influence values of the pixel and the gray difference of the pixel at the same position in the target frame and the adjacent image frame, including:

in a preset second range corresponding to each pixel point to be denoised, acquiring the average value of the gray value difference of the pixel point at the same position in the target frame and the adjacent image frame of each pixel point as a difference index;

Taking other pixel points except the central pixel point in each preset second range as analysis points, and obtaining a similar index of each analysis point and the corresponding central pixel point according to the difference of the illumination influence value and the difference index between each analysis point and the corresponding central pixel point, wherein the difference of the illumination influence value and the difference index are in negative correlation with the similar index;

and taking the analysis point with the similarity index larger than the preset similarity threshold value as a reference point of the corresponding pixel point to be denoised.

Further, in the preset neighborhood centered on each reference point, obtaining a noise representation value of the reference point according to the gray scale difference between each reference point and all neighborhood pixel points, including:

and normalizing the difference between the gray value of each reference point and the gray value average value of all the neighborhood pixel points to obtain a noise representation value of each reference point.

Further, the obtaining the gray value weight of each reference point according to the noise representation value of the reference point, the position distribution of the reference point and the corresponding pixel point to be denoised, and the gray value variation condition of the pixel point of the reference point at the same position in the target frame and the adjacent image frame, includes:

Taking the distance between each reference point and the corresponding pixel point to be denoised as a distance factor;

when the noise representation value of the reference point is smaller than or equal to a preset representation threshold value, taking the reference point as a first type reference point, taking the difference between the gray value of each first type reference point and the gray value of a pixel point at the same position in any one image frame adjacent to the target frame as a molecule, taking the sum value of the product of the distance factor corresponding to each first type reference point and the noise representation value and a preset third parameter as a denominator, and taking the value obtained by normalizing the obtained ratio as the gray value weight of each first type reference point;

when the noise representation value of the reference point is larger than the preset representation threshold value, taking the reference point as a second class reference point, obtaining the sum value of the difference between the gray value of each second class reference point and the gray value of the pixel point at the same position in two image frames closest to the target frame, taking the sum value of the product of the distance factor corresponding to each second class reference point, the noise representation value and the preset time interval and the sum value of the preset fourth parameter as a denominator, and taking the normalized value of the obtained ratio as the gray value weight of each second class reference point.

Further, the updating the gray value of each pixel to be denoised according to the gray value weights of all the reference points corresponding to each pixel to be denoised and the gray values of the pixel of the reference point at the same position in the target frame and the adjacent image frame, and obtaining and transmitting the enhanced image includes:

taking the updated value of the gray value of each pixel to be denoised as an updated gray value, wherein the formula model of the updated gray value of each pixel to be denoised is as follows:

wherein,indicate->Updating gray value of each pixel to be denoised, < >>Indicate->Total number of first type reference points of the pixel points to be denoised, < >>Representation ofFirst->Total number of second class reference points of the pixel points to be denoised, < >>Indicate->The +.>Gray value weights of the first class of reference points,>indicate->The +.>Gray values of the first type of reference points in the target frame +.>Indicate->The +.>Gray value weights of the second class of reference points,>indicate->The +.>The gray value average value of the pixel points of the second class reference points at the same position in two image frames closest to the target frame;

And replacing the original gray value with the updated gray value of each pixel point to be denoised in the target frame to obtain an enhanced image, and transmitting the enhanced image.

Further, the step of screening suspected denoising pixels in the target frame according to the gray level change of pixels at the same position in the target frame and the adjacent image frame includes:

calculating the average value of the gray value difference between each pixel point in the target frame and the pixel point at the same position in the adjacent image frame to obtain a judgment index;

and taking the pixel point with the judgment index larger than the preset judgment threshold value in the target frame as a suspected denoising pixel point.

Further, the acquiring the gray gradient direction of the edge line and each pixel point in each image frame includes:

and acquiring edge lines in each image frame and the gray gradient direction of each pixel point based on the Canny operator.

The invention also provides a video image transmission quality enhancement system, which comprises:

a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of any one of the methods when the computer program is executed.

The invention has the following beneficial effects:

Because the difference between two adjacent image frames with shorter time interval is smaller and the change between the images is not obvious, the invention sets the preset time interval to acquire the image frames, initially improves the denoising accuracy, and then acquires the edge line in each image frame and the gray gradient direction of each pixel point so as to prepare for the subsequent analysis process; in order to improve the image enhancement efficiency, the method acquires the changed local area in the video image frame, namely screens out pixel points to be denoised, however, before screening out the pixel points to be denoised, the importance of the local area change needs to be analyzed, because the change degree of the local area change is small, and the change degree of the local area change is possibly caused by the movement of light shadow generated by illumination on an object and does not belong to the active change of the object, the importance degree of denoising is not high, so that suspected denoising pixel points can be screened out according to the gray level change of the pixel points in adjacent frames, further in a preset first range, the illumination influence value of the pixel points is obtained according to the gray level difference between the pixel points on a straight line where the gray level gradient direction of the pixel points is located, and then the pixel points to be denoised are screened out in the suspected denoising pixel points by combining the positions of edge lines; further, a reference point corresponding to each pixel to be denoised is required to be selected, and when the reference point is selected, a pixel similar to the pixel to be denoised is required to be selected as the reference point, so that the denoising effect obtained by the method is better, and the difference of illumination influence values of the pixels and the gray level difference in the adjacent image frames are analyzed in a preset second range with the pixel to be denoised as the center, and the reference points are screened; the reference point is also affected by noise, so that the noise representation value of the reference point needs to be judged, the gray value weight which is finally occupied by the reference point is determined according to the noise representation value of the reference point, the gray value change condition of the reference point in the adjacent image frames and the position relation of the reference point and the corresponding pixel point to be denoised, and finally the pixel point to be denoised can be denoised, namely the gray value is updated according to the gray value of the reference point corresponding to each pixel point to be denoised in the image frames and the gray value weight corresponding to the reference point, so that the enhanced image is transmitted. According to the method, the pixel points to be denoised are screened out by analyzing the gray level change condition and the like of the pixel points in the adjacent image frames at the preset time interval, then the gray level difference between the pixel points in the preset second range corresponding to the pixel points to be denoised and the pixel points are analyzed, and representative reference points are screened out, so that the reference points are analyzed, gray level weight of the reference points is obtained, and then the pixel points to be denoised are denoised, so that local denoising is realized, the denoising effect is ensured, and meanwhile, the image enhancement efficiency is improved.

Drawings

In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.

Fig. 1 is a flowchart of a video image transmission quality enhancement method according to an embodiment of the present invention.

Detailed Description

In order to further describe the technical means and effects adopted by the present invention to achieve the preset purposes, the following detailed description refers to specific embodiments, structures, features and effects of a video image transmission quality enhancement method and system according to the present invention with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

The following specifically describes a specific scheme of a video image transmission quality enhancement method and system provided by the present invention with reference to the accompanying drawings.

Referring to fig. 1, a method flowchart of a video image transmission quality enhancement method according to an embodiment of the present invention is shown, the method includes the following steps:

step S1: and acquiring image frames based on a preset time interval, and acquiring an edge line in each image frame and a gray gradient direction of each pixel point.

In the process of image generation, transmission or transformation, the image is influenced by a light source, an imaging system, channel bandwidth, noise and other factors, and degradation phenomena such as low contrast, insufficient dynamic range, reduced definition, obvious noise and the like can occur; therefore, image enhancement is required to improve these problems. However, when the image frames in the video are acquired, because the differences between the image frames are often small, that is, the changes between the image frames are not obvious, the preset time interval is set in the embodiment of the invention to acquire the image frames in the video, so that the aim of preliminarily improving the denoising accuracy and the denoising efficiency is achieved. It should be noted that, in this embodiment of the present invention, the video frame rate is 30fps, and 30 frames of images are displayed per second, that is, the time difference between two frames of images is about 0.033 seconds, so in order to avoid missing the time node for analyzing the pixel change, the preset time interval should not be too large, so the preset time interval is set to be about 0.066 seconds, that is, the 1 st frame, the 3 rd frame, the 5 th frame, and other images are extracted as the image frames in the embodiment of the present invention.

In image processing, edges are boundaries between different regions in an image and thus are important features in an image, so that the edge lines in each image frame and the gray gradient direction of each pixel point can be obtained for use in subsequent analysis.

Preferably, in one embodiment of the present invention, acquiring a gray gradient direction of an edge line and each pixel point in each image frame includes:

and acquiring edge lines in each image frame and the gray gradient direction of each pixel point based on the Canny operator. It should be noted that the Canny operator is a technical means well known to those skilled in the art, and will not be described herein.

Step S2: optionally selecting one of the image frames as a target frame; in a target frame, in a preset first range with each pixel point as a center, obtaining an illumination influence value of the center pixel point according to gray level differences among the pixel points on a straight line where gray level gradient directions of all the pixel points are located; screening suspected denoising pixel points in the target frame according to the gray level change of the pixel points at the same position in the target frame and the adjacent image frames; and obtaining the pixel points to be denoised according to the preset time interval, the position distribution of each suspected denoising pixel point and the edge line and the illumination influence value of each suspected denoising pixel point.

When denoising an image frame, in order to improve denoising efficiency, pixel points which change in the image frame are analyzed to denoise the image frame, but in the screening process, the change of the pixel points can be divided into active change caused by object movement and passive change caused by movement of a light shadow on the object due to external factors such as sun movement, illumination change, and the denoising importance of the passive change is lower than that of the active change. For convenience of explanation and explanation, in the embodiment of the present invention, any one of all image frames is taken as a target frame, and the whole process of the present invention is explained by analyzing the target frame.

Preferably, in one embodiment of the present invention, in a target frame, in a preset first range centered on each pixel, obtaining an illumination influence value of the center pixel according to a gray scale difference between pixels on a line where gray scale gradient directions of all pixels are located, including:

Firstly, setting a preset first range by taking each pixel point as a center, and calculating illumination influence factors of the pixels in the preset first range corresponding to each pixel point, wherein the specific calculation method comprises the following steps: the gray gradient direction can often represent the gray variation of the pixel points, namely, the gradual change condition caused by illumination can be reflected, so that on the straight line where the gray gradient direction of each pixel point is positioned, each preset number of pixel points on the left side and the right side of each pixel point are taken as target points; and then calculating accumulated values of gray value differences between every two target points of each pixel point to obtain illumination influence factors of each pixel point.

And finally, taking other pixel points except the central pixel point in each preset first range as comparison points of the central pixel point, and carrying out negative correlation mapping and normalization on the value obtained by accumulating the differences of the illumination influence factors between all the comparison points and the corresponding central pixel points to obtain the illumination influence value of the central pixel point. The formula model of the illumination influence value of each pixel point can be specifically, for example:

wherein,indicate->Illumination influence value of each pixel point, +.>Indicate->The total number of contrast points for each pixel point,indicate- >Illumination influence factor of each pixel point, +.>Indicate->The +.>Illumination influencing factors of individual contrast spots, +.>Expressed as natural constant->An exponential function of the base.

In a formula model of illumination influence values, firstly, accumulated values of gray value differences between every two target points of each pixel point in a preset first range of each pixel point are obtained, namely illumination influence factors, wherein the smaller the illumination influence factors are, the more likely the gray values between the target points are similar, namely the greater the possibility of representing gradual change characteristics is, namely the possibility of being influenced by illumination is; then each preset first range except the central pixel pointThe other pixel points of the (a) are taken as the contrast points of the central pixel point, the differences of the illumination influence factors of all the contrast points and the central pixel point are analyzed and accumulated, and at the moment, the differences of the illumination influence factors of the contrast points and the central pixel point are differentThe smaller the first range, the more similar the change of the central pixel point and the contrast points is, namely the more the possibility of being influenced by illumination, so the central pixel point and all corresponding contrast points are treated as +.>And carrying out negative correlation mapping and normalization on the accumulated values to realize logic relation correction and obtain illumination influence values, wherein the larger the illumination influence values are, the more likely the change of the pixel points is caused by illumination.

It should be noted that, in this embodiment of the present invention, the preset first range is: the over-center pixel point is used for making a perpendicular line of a straight line where the gradient direction is located, three pixel points on two sides of the center pixel point on the perpendicular line, namely, a preset first range of each pixel point totally comprises 7 pixel points, and a size implementation person arranged in a specific range can be adjusted according to an implementation scene and is not limited.

After the illumination influence of each pixel point is obtained, the gray value change condition of the pixel point is required to be analyzed, if the gray value change condition is gentle in the adjacent image frames, the gray value change condition is considered to contain less useful information, the denoising necessity is not great, if the gray value change condition is severe, the gray value change condition is considered to contain more useful information, denoising is required to be carried out, the gray value change condition is considered to be a suspected denoising pixel point, and preparation is carried out for the subsequent screening of the pixel point to be denoised.

Preferably, in one embodiment of the present invention, the step of selecting suspected denoising pixels in a target frame according to gray scale variation of pixels at the same position in the target frame and adjacent image frames includes:

the gray value change of the pixel points in the adjacent image frames can reflect the motion and change of an object in the image, the pixel points with more severe change can be screened out through analyzing the change, and the pixel points are used as suspected denoising pixel points needing denoising, so that the average value of the gray value difference of each pixel point in the target frame and the pixel points at the same position in the adjacent image frames is calculated, a judgment index is obtained, the judgment index can represent the gray change condition of the pixel points in the adjacent image frames, and the larger the judgment index of the pixel points is, the more severe the gray change condition of the pixel points is, the more active change is likely to happen to the object instead of the gray change caused by illumination influence. And finally, taking the pixel point with the judgment index larger than the preset judgment threshold value in the target frame as a suspected denoising pixel point. It should be noted that, the preset judgment threshold is set to 5, and the specific numerical value implementer can adjust according to the implementation scenario, which is not limited herein; in the process of calculating the judging index, if the target frame is the first frame or the last frame, only one adjacent image frame exists, and if the target frame is the image frame except the first frame or the last frame, two adjacent image frames exist.

After the suspected denoising pixel points are screened according to the gray level change condition of the pixel points in the adjacent image frames and the illumination influence value of the suspected denoising pixel points is obtained, the suspected denoising pixel points can be used as indexes for screening the pixel points to be denoised, and the edge lines are important characteristics in the image because the edge lines are boundary lines between different areas in the image, so that the edge lines can also be used as one of indexes for screening the pixel points to be denoised and are combined with a preset time interval to obtain the pixel points to be denoised.

Preferably, in one embodiment of the present invention, obtaining the pixel to be denoised according to a preset time interval, a position distribution of each suspected denoised pixel and an edge line, and an illumination influence value of each suspected denoised pixel includes:

when the suspected denoising pixel points are closer to the edge line, the suspected denoising pixel points are more likely to contain useful information, so that denoising is more needed for improving the image quality, and the shortest distance between each suspected denoising pixel point and the edge line closest to the suspected denoising pixel point is firstly obtained and is used as a distance parameter of each suspected denoising pixel point; if the edge line is a regular straight line, the calculation method is a point-to-straight line distance, and if the edge line is an irregular curve, the distance between the suspected denoising pixel point and each pixel point on the edge line can be calculated, and the minimum value in the distance is taken as the shortest distance, namely a distance parameter.

Then, the denoising possibility of each suspected denoising pixel point can be obtained according to the distance parameter, the illumination influence value and the preset time interval of each suspected denoising pixel point, the denoising possibility is positively correlated with the time interval, the illumination influence value and the distance parameter are negatively correlated with the denoising possibility, and the denoising possibility represents the importance degree of the suspected denoising pixel point; and finally, the suspected denoising pixel point with the denoising possibility larger than the preset denoising threshold value can be used as the pixel point to be denoised. The formula model of the denoising possibility may specifically be, for example:

wherein,indicate->Denoising possibility of each suspected denoising pixel point,/->Indicate->Distance parameter of suspected denoising pixel point, < ->Indicate->Illumination influence value of each suspected denoising pixel,/->Representing a preset time interval,/->Representing a preset first parameter, ">Representing the normalization function.

In the formula model of denoising possibility, when the preset time interval is larger, the pixel point is considered to be obviously changed, and the denoising is performed so that the reliability of the image quality is improved, so that the pixel point is used as one of indexes for obtaining the denoising possibility; based on the above analysis, it can be seen that: when the suspected denoising pixel point is closer to the edge line, namely, the distance parameter is smaller, the suspected denoising pixel point is more likely to contain useful information, so that denoising is more likely to be performed on the suspected denoising pixel point to improve the image quality, and thus useful information is obtained; therefore, taking the sum of the product of the illumination influence value and the distance parameter and the preset first parameter as a denominator, taking the preset time interval as a numerator, when the denominator is smaller and the numerator is larger, the pixel point is required to be denoised, the image quality is improved, useful information is acquired, and the obtained ratio, namely the denoising possibility is higher. It should be noted that, in this embodiment of the present invention, the preset denoising threshold value is set to 0.2; presetting a first parameter The value is set to 0.001, which is used for preventing the denominator from being 0, and the specific numerical values can be adjusted according to the implementation scene, and the method is not limited herein.

Therefore, the pixel points to be denoised in the target frame can be screened out, and only the pixel points to be denoised are denoised in the subsequent analysis process, so that local image enhancement is realized, and the image enhancement efficiency is effectively improved.

Step S3: screening reference points according to the difference of illumination influence values of the pixel points and the gray level difference of the pixel points at the same position in the target frame and the adjacent image frames in a preset second range taking each pixel point to be denoised as a center; in a preset neighborhood taking each reference point as a center, obtaining a noise representation value of the reference point according to the gray level difference between each reference point and all neighborhood pixel points; and obtaining the gray value weight of each reference point according to the noise representation value of the reference point, the position distribution of the reference point and the corresponding pixel point to be denoised and the gray value change condition of the pixel point of the reference point at the same position in the target frame and the adjacent image frame.

Based on the step S2, pixel points to be denoised in the target frame can be screened, and local denoising and enhancement of the image are realized by denoising the pixel points to be denoised, so that the image enhancement efficiency is improved; the embodiment of the invention screens the similar reference points from the preset second range corresponding to each pixel point to be denoised based on the idea of non-local mean filtering, thereby realizing the enhancement of the pixel point to be denoised by analyzing the reference points.

Preferably, in one embodiment of the present invention, in a preset second range centered on each pixel to be denoised, screening a reference point according to a difference in illumination influence values of the pixel and a gray difference of the pixel at the same position in the target frame and the adjacent image frame, includes:

the filtering of the reference points influences the denoising effect of the pixel points to be denoised, so that when the reference points are selected, the more similar the selected reference points are to the pixel points to be denoised, the more reliable the result obtained after denoising the pixel points to be denoised according to the reference points is.

Since the pixel points to be denoised are essentially the pixel points with gray values changed in the adjacent image frames, the reference points should also select the pixel points with gray values changed, so that the average value of the gray value difference between the gray value of each pixel point and the gray value difference between the pixel points at the same position in the image frame adjacent to the target frame is obtained as a difference index in the preset second range corresponding to each pixel point to be denoised.

Then taking other pixel points except the central pixel point in each preset second range as analysis points, and calculating the similarity between the analysis points and the central pixel points: and obtaining a similar index of each analysis point and the corresponding central pixel point according to the difference of the illumination influence value and the difference index between each analysis point and the corresponding central pixel point, wherein the difference of the illumination influence value and the difference index are in negative correlation with the similar index. And finally, taking the analysis point with the similarity index larger than the preset similarity threshold value as a reference point of the corresponding pixel point to be denoised. The formula model of the similarity index may specifically be, for example:

Wherein,indicate->Pixel point to be denoised and the +.>Similarity index of individual reference points,/>Indicate->Illumination influence values of pixels to be denoised,/->Indicate->The +.>Illumination influence value of individual reference points, +.>Indicate->Difference index of pixel points to be denoised, < >>Indicate->The +.>Difference index of individual reference points,>representing a preset second parameter, ">Representing the normalization function.

In the formula model of the similarity index, in the preset second range centered on each pixel to be denoised, since the illumination influence value in step S2 can reflect the gray level change condition of a certain pixel on the straight line where the gradient direction is located, the illumination influence value can be used as one of the indexes of the screening reference point, when the difference between the illumination influence values of the pixel in the preset second range and the center pixel is smaller, namelyThe smaller the pixel point is, the more similar the pixel point is to the central pixel point, and the more reliable the result is obtained by taking the pixel point as a reference point of the central pixel point to carry out a subsequent analysis process; similarly, the difference index is an index obtained according to the gray level change of the pixel point in the adjacent image frames, so when the difference between the difference index of the pixel point in the preset second range and the difference index of the center pixel point is more approximate, namely ∈ - >The smaller the time, the more similar the change condition of the pixel point and the central pixel point is, the more reliable the pixel point is used as a reference point; so combining the difference of the illumination influence value and the difference of the difference index, multiplying the two, taking the value obtained by adding the difference of the illumination influence value and the preset second parameter as a denominator to realize logic relation correction, and normalizing the obtained ratio, wherein the difference of the illumination influence value is->And difference index->The smaller the pixel is, the smaller the denominator is, and the larger the similarity index is, which indicates that the pixel point is similar to the center pixel point.

It should be noted that, in this embodiment of the present invention, the preset second range is 101×101, and the preset similarity threshold is set to 0.6; presetting a second parameterThe value of 0.001 is used for preventing the denominator from being 0, and the specific range size and the numerical value size can be adjusted according to the implementation scene, and the method is not limited.

Therefore, the reference point of each pixel point to be denoised can be screened out, and then the reference point of each pixel point to be denoised can be analyzed to realize the denoising process of the pixel point to be denoised.

According to the embodiment of the invention, the reference point of each pixel point to be denoised is screened, the gray value of the pixel point to be denoised is determined through the gray value of the reference point, so that the updated gray value of the pixel point to be denoised is more stable, the influence of noise is avoided, the visual effect of an image is improved, but the reference point is possibly influenced by the noise due to higher similarity between the reference point and the corresponding pixel point to be denoised, and the influence degree of the reference point is possibly different, and the required processing method is also correspondingly changed, so that the influence degree of the reference point by the noise, namely the noise representation value of the reference point, is needed to be judged first.

Preferably, in one embodiment of the present invention, in a preset neighborhood centered on each reference point, obtaining a noise representation value of the reference point according to a gray difference between each reference point and all neighborhood pixel points includes:

if the influence degree of noise on a certain reference point is large, the gray values of the reference point and surrounding pixel points should have large difference, otherwise, if the influence degree of noise on a certain reference point is small, the difference of the gray values of the reference point and surrounding pixel points should be small, so that a preset neighborhood can be set by taking each reference point as a center, and then the average value of the gray values of all neighborhood pixel points and the difference of the gray values of corresponding reference points are normalized to obtain the noise representation value of each reference point. The formula model of the noise representation value is:

wherein,indicate->Noise representation value of the individual reference points, +.>Indicate->Gray values of individual reference points +.>Indicate->First->Gray value of each neighborhood pixel, +.>Indicate->The total number of neighborhood pixels for each reference point,representing the normalization function.

In the formula model of the noise expression value, all neighborhood images corresponding to the reference points are firstly obtainedThe mean value of the gray values of the pixels, i.e The value can be used for representing the average level of the gray values of the pixel points in the local area where the reference point is located, and then the gray values of the reference point are compared with the average value, namely +.>The smaller the value, the smaller the average level difference between the gray values of the reference point and all the neighborhood pixel points is, the less the reference point is affected by noise, that is, the noise expression value of the reference point is considered to be small, whereas the larger the value is, the larger the average level difference between the gray values of the reference point and all the neighborhood pixel points is, the greater the reference point is affected by noise, that is, the noise expression value of the reference point is considered to be large. It should be noted that, in this embodiment of the present invention, the preset neighborhood is set to 5×5, and the specific numerical value implementation may be adjusted according to the implementation scenario, which is not limited herein.

Thus, the noise representation value of the reference point can be obtained, and in order to better balance the local and global characteristics of the image and improve the quality and definition of the image, the gray value weight of each reference point is set and then weighted in the embodiment of the invention, so that the pixel point to be denoised is denoised. The gray level change can be regarded as a certain change of an object or a scene represented by the pixel point in the image, and can be used as one of indexes for measuring the gray level weight, the noise representation value of the reference point can also influence the setting of the gray level weight, and meanwhile, the position relationship between the reference point and the pixel point to be denoised can also influence the setting of the gray level weight, so that the gray level weight of each reference point is obtained by combining the gray level change.

Preferably, in one embodiment of the present invention, obtaining the gray value weight of each reference point according to the noise representation value of the reference point, the position distribution of the reference point and the corresponding pixel point to be denoised, and the gray value variation situation of the pixel point of the reference point at the same position in the target frame and the adjacent image frame, includes:

the distance factor can be considered for the position distribution of the reference points and the corresponding pixel points to be denoised, so that the distance between each reference point and the corresponding pixel points to be denoised is taken as the distance factor. Then, since the noise influence degree of each reference point is different, the noise representation value of the reference point of each pixel point to be denoised is compared with a preset representation threshold value in order to realize the operation of classifying the reference points and then acquiring the gray value weight, in order to realize the follow-up result more accurate, the specific process is as follows.

When the noise representation value of the reference point is smaller than or equal to the preset representation threshold value, the noise representation value of the reference point is taken as a first type reference point, and the noise influence degree of the first type reference point is small at the moment, so that the gray level change of the reference point in two continuous frames can be analyzed to better capture the real change of an object or a scene, the difference between the gray level value of each first type reference point and the gray level value of a pixel point at the same position in any image frame adjacent to the target frame is taken as a molecule, the sum value of the distance factor corresponding to each first type reference point and the noise representation value and the preset third parameter is taken as a denominator, and the value obtained by normalizing the ratio is taken as the gray level weight of each first type reference point.

When the noise representation value of the reference point is larger than the preset representation threshold, the reference point is used as a second type reference point, at the moment, the influence degree of the noise on the second type reference point is larger, the gray level change condition of the reference point in the three-frame image is analyzed to be more reliable, the three-frame image is analyzed to enable the gray level change condition of the reference point to be more obvious and stable, the subsequent analysis is facilitated, therefore, the sum value of the difference between the gray level value of each second type reference point and the gray level value of the pixel point at the same position in two image frames closest to the target frame is obtained respectively and is used as a molecule, the sum value of the product of the distance factor corresponding to each second type reference point, the noise representation value and the preset time interval and the sum value of the preset fourth parameter is used as a denominator, and the value after the normalization of the ratio is used as the gray level weight of each second type reference point.

Reference points of the first kindFor example, the formula model of the gray value weight is:

wherein,representing a first type of reference point->Gray value weight of ∈10->Representing a first type of reference point->Is a difference between the gray value of the pixel point at the same position in any one of the image frames adjacent to the target frame,/and the gray value of the pixel point at the same position in the image frame adjacent to the target frame>Representing a first type of reference point- >Corresponding distance factor,/->Representing a first type of reference point->Noise representation value of +.>Indicating that the third parameter is preset and that,representing the normalization function.

In the formula model of the gray value weight of the first type of reference points, the gray value of the first type of reference points in the target frame and any adjacent image frame are calculated because the first type of reference points are slightly affected by noiseDifference in gray values at the same positionThe condition that the gray value of the reference point changes can be characterized, if the larger the value is, the more information possibly contained is indicated, the larger the gray value weight of the reference point is, so the gray value weight of the reference point is taken as a molecule; for distance factor->When the value of the distance factor is smaller, the closer the distance between the reference point and the corresponding pixel point to be denoised is, the more reliable the result obtained by updating the gray value of the pixel point to be denoised by using the gray value of the reference point is, so the larger the gray value weight of the reference point is, and when the noise representation value of the reference point is smaller, the more reliable the gray value weight obtained by participating the reference point in the calculation is, so the distance factor and the noise representation value are combined, and the multiplied value is taken as a denominator; at this time, the smaller the denominator, the larger the numerator, the more reliable the result obtained when the reference point participates in the subsequent updating process of the gray value of the pixel point to be denoised, namely the larger the gray value weight.

Reference points of the second kindFor example, the formula model of the gray value weight is:

wherein,representing reference points of the second kind->Gray value weight of ∈10->Representing reference points of the second kind->Sum of differences between gray values of pixels at the same position in two image frames closest to the target frame and gray values of pixels at the same position in the target frame,/>Representing reference points of the second kind->Corresponding distance factor,/->Representing reference points of the second kind->Noise representation value of +.>Representing a preset time interval,/->Representing a preset fourth parameter, ">Representing the normalization function.

In the formula model of the gray value weight of the second class of reference points, since the second class of reference points are greatly influenced by noise, the gray change condition of the second class of reference points in three frames of images is more reliable to analyze, and therefore the sum value of the difference between the gray value of the second class of reference points in the target frame and the gray value of the reference points in the same position in two image frames closest to the target frame is respectively obtainedThe larger the sum value, the more information that may be contained, and the larger the gray value weight of the reference point should be, so the reference point is taken as a molecule; for distance factor->When the value of the distance factor is smaller, the reference point and the corresponding pixel point to be denoised are described The closer the distance is, the more reliable the result obtained by updating the gray value of the pixel to be denoised by using the gray value of the reference point is, so the larger the gray value weight of the reference point should be, and likewise, the more reliable the gray value weight obtained by taking the reference point into account when the noise representation value of the reference point is smaller is, so the distance factor, the noise representation value and the value multiplied by combining the preset time interval are taken as denominators; at this time, a->The gray level change rate of the gray level value of the reference point in the adjacent image frames can be represented, and the larger the gray level value is, the larger the gray level value change of the reference point is, and the larger the weight is; in summary, the smaller the denominator, the larger the numerator, the more reliable the result obtained when the reference point participates in the subsequent updating process of the gray value of the pixel point to be denoised, namely the larger the gray value weight.

It should be noted that, the preset performance threshold is set to 0.5; presetting a third parameterAnd preset fourth parameter->The function is to prevent the denominator from being 0, the value is set to be 0.001, the specific value can be adjusted according to the implementation scene, and the method is not limited; when the gray value weight of the first type of reference point is calculated, selecting one image frame adjacent to the target frame, in the embodiment of the invention, the previous image frame adjacent to the target frame can be uniformly selected, and when the first image frame is the target frame, the next image frame adjacent to the first image frame is selected; the specific examples are not limited.

So far, the gray value weight corresponding to each reference point can be obtained, and then the gray value of the pixel point to be denoised corresponding to the reference point can be updated according to the gray value weight and the gray value.

Step S4: and updating the gray value of each pixel point to be denoised according to the gray value weights of all the reference points corresponding to each pixel point to be denoised and the gray values of the pixel points of the reference points at the same position in the target frame and the adjacent image frames, so as to obtain and transmit the enhanced image.

In the step, the pixel points to be denoised are screened out, and the reference point corresponding to each pixel point to be denoised and the gray value weight of the reference point are obtained, so that the gray value of the pixel points to be denoised can be determined according to the gray value weight and the gray value of the reference point, thereby avoiding the influence of noise and improving the definition and visual effect of the image.

Preferably, in one embodiment of the present invention, according to gray value weights of all reference points corresponding to each pixel to be denoised and gray values of pixel points of the reference points at the same position in the target frame and the adjacent image frame, updating the gray value of each pixel to be denoised, obtaining and transmitting an enhanced image, including:

Firstly, taking the updated gray value of each pixel to be denoised as an updated gray value, and then the formula model of the updated gray value of each pixel to be denoised is as follows:

wherein,indicate->Updating gray value of each pixel to be denoised, < >>Indicate->Total number of first type reference points of the pixel points to be denoised, < >>Indicate->Second class reference of pixel points to be denoisedTotal number of points>Indicate->The +.>Gray value weights of the first class of reference points,>indicate->The +.>Gray values of the first type of reference points in the target frame +.>Indicate->The +.>Gray value weights of the second class of reference points,>indicate->The +.>The second type of reference points are the gray value averages of the pixel points at the same positions in the image frames adjacent to the target frame.

In the formula model for updating the gray value, the gray value weight of each first type of reference point of each pixel point to be denoised is normalized, namely, each pixel point to be denoised is obtainedThe duty ratio of the gray value weight of each first type of reference point to the sum of the gray value weights of all first type of reference points is obtained Then multiplying the value with the gray value of the first type reference point in the target frame to obtain gray factor of each first type reference point>The method comprises the steps of carrying out a first treatment on the surface of the Similarly, the gray value weights of the second class reference points of the pixel points to be denoised are normalized to obtain the duty ratio of the gray value weights of the second class reference points of the pixel points to be denoised in the sum of the gray value weights of all the second class reference points, thereby obtainingSince the method of acquiring the gray value weight of the second type of reference point analyzes the difference of gray values of the reference point at the same position in the target frame as in the two image frames closest to the target frame, here too, the gray value mean +_of the second type of reference point at the same position in the two image frames closest to the target frame is acquired>Then the value is combined with +.>Multiplying to obtain the gray factor of each second class reference point; and then, the gray factors of all the reference points of the pixel points to be denoised can be added to obtain the updated gray value of the pixel points to be denoised.

And finally, replacing the original gray value with the updated gray value of each pixel point to be denoised in the target frame, so that an enhanced image can be obtained, and the enhanced image is transmitted.

According to the embodiment of the invention, the pixel points to be denoised are screened out, so that local denoising is realized, the reference points are screened, the gray values of the pixel points to be denoised are updated according to the reference points, the denoising effect is ensured, and the image enhancement efficiency is improved while the image quality is improved.

In summary, since the difference between two adjacent image frames with shorter time interval is smaller and the change between the images is not obvious, the invention sets the preset time interval to acquire the image frames, initially improves the denoising accuracy, and then acquires the edge line in each image frame and the gray gradient direction of each pixel point to prepare for the subsequent analysis process; in order to improve the image enhancement efficiency, the method acquires the changed local area in the video image frame, namely screens out pixel points to be denoised, however, before screening out the pixel points to be denoised, the importance of the local area change needs to be analyzed, because the change degree of the local area change is small, and the change degree of the local area change is possibly caused by the movement of light shadow generated by illumination on an object and does not belong to the active change of the object, the importance degree of denoising is not high, so that suspected denoising pixel points can be screened out according to the gray level change of the pixel points in adjacent frames, further in a preset first range, the illumination influence value of the pixel points is obtained according to the gray level difference between the pixel points on a straight line where the gray level gradient direction of the pixel points is located, and then the pixel points to be denoised are screened out in the suspected denoising pixel points by combining the positions of edge lines; further, a reference point corresponding to each pixel to be denoised is required to be selected, and when the reference point is selected, a pixel similar to the pixel to be denoised is required to be selected as the reference point, so that the denoising effect obtained by the method is better, and the difference of illumination influence values of the pixels and the gray level difference in the adjacent image frames are analyzed in a preset second range with the pixel to be denoised as the center, and the reference points are screened; the reference point is also affected by noise, so that the noise representation value of the reference point needs to be judged, the gray value weight which is finally occupied by the reference point is determined according to the noise representation value of the reference point, the gray value change condition in the adjacent image frames of the reference point and the position relation between the reference point and the corresponding pixel point to be denoised, and finally, the pixel point to be denoised is denoised, namely the gray value is updated according to the gray value of the reference point corresponding to each pixel point to be denoised in the image frames and the gray value weight corresponding to the reference point, so that the enhanced image is transmitted. According to the method, the pixel points to be denoised are screened out by analyzing the gray level change condition and the like of the pixel points in the adjacent image frames at the preset time interval, then the gray level difference between the pixel points in the preset second range corresponding to the pixel points to be denoised and the pixel points are analyzed, and representative reference points are screened out, so that the reference points are analyzed, gray level weight of the reference points is obtained, and then the pixel points to be denoised are denoised, so that local denoising is realized, the denoising effect is ensured, and meanwhile, the image enhancement efficiency is improved.

An embodiment of the present invention also provides a video image transmission quality enhancement system, which includes a memory, a processor, and a computer program, where the memory is configured to store a corresponding computer program, and the processor is configured to execute the corresponding computer program, and the computer program is capable of implementing the steps of a video image transmission quality enhancement method when executed on the processor.

It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.

Claims

1. A method for video image transmission quality enhancement, the method comprising:

2. The method for enhancing video image transmission quality according to claim 1, wherein the obtaining the illumination influence value of the center pixel according to the gray scale difference between the pixels on the straight line where the gray scale gradient direction of all the pixels is located within the preset first range with each pixel as a center comprises:

3. The method for enhancing video image transmission quality according to claim 1, wherein the obtaining the pixel to be denoised according to the preset time interval, the position distribution of each suspected denoised pixel and the edge line, and the illumination influence value of each suspected denoised pixel comprises:

4. The method according to claim 1, wherein the screening the reference points according to the difference of the illumination influence values of the pixel points and the gray level difference of the pixel points at the same position in the target frame and the adjacent image frame in the preset second range with each pixel point to be denoised as a center comprises:

5. The method according to claim 1, wherein the obtaining the noise representation value of the reference point in the preset neighborhood centered on each reference point according to the gray scale difference between each reference point and all neighborhood pixel points comprises:

6. The method according to claim 1, wherein the obtaining the gray value weight of each reference point according to the noise representation value of the reference point, the position distribution of the reference point and the corresponding pixel point to be denoised, and the gray value variation of the pixel point of the reference point at the same position in the target frame and the adjacent image frame comprises:

7. The method for enhancing transmission quality of video image according to claim 6, wherein updating the gray value of each pixel to be denoised according to the gray value weights of all the reference points corresponding to each pixel to be denoised and the gray values of the pixel of the reference point at the same position in the target frame and the adjacent image frame, and obtaining and transmitting the enhanced image comprises:

wherein,indicate->Updating gray value of each pixel to be denoised, < >>Indicate->Total number of first type reference points of the pixel points to be denoised, < >>Indicate->Second class of parameters of pixel points to be denoisedTotal number of points>Indicate->The +.>Gray value weights of the first class of reference points,>indicate->The +.>Gray values of the first type of reference points in the target frame +.>Indicate->The +.>Gray value weights of the second class of reference points,>indicate->The +.>The gray value average value of the pixel points of the second class reference points at the same position in two image frames closest to the target frame;

8. The method according to claim 1, wherein the step of screening the suspected denoising pixels in the target frame according to the gray level change of the pixels at the same position in the target frame and the adjacent image frame comprises:

9. The method according to claim 1, wherein the acquiring the gray gradient direction of the edge line and each pixel in each image frame comprises:

10. A video image transmission quality enhancement system comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any one of claims 1-9 when the computer program is executed by the processor.