CN115063326A

CN115063326A - Infrared night vision image efficient communication method based on image compression

Info

Publication number: CN115063326A
Application number: CN202210989581.0A
Authority: CN
Inventors: 符士华
Original assignee: Weihai Tiantuo Hechuang Electronic Engineering Co ltd
Current assignee: Weihai Tiantuo Hechuang Electronic Engineering Co ltd
Priority date: 2022-08-18
Filing date: 2022-08-18
Publication date: 2022-09-16
Anticipated expiration: 2042-08-18
Also published as: CN115063326B

Abstract

The invention relates to the technical field of image data processing, in particular to an efficient communication method of an infrared night vision image based on image compression, which comprises the steps of obtaining an N-frame infrared night vision image to be transmitted, and carrying out image processing on the N-frame infrared night vision image to be transmitted to obtain a target area image to be reconstructed; determining information content indexes corresponding to bit layer images of a second target area image to be reconstructed, and further determining weight values corresponding to the bit layer images; and determining a weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed, and further determining a weight value of each bit layer image of each target area image to be reconstructed, so that compression transmission of each target area image to be reconstructed after reconstruction processing is realized. The invention solves the problem of poor definition of detail information of the transmitted compressed image, improves the image detail and the transmission efficiency of the compressed image, and is applied to the related field of compressed image transmission.

Description

Infrared night vision image efficient communication method based on image compression

Technical Field

The invention relates to the technical field of image data processing, in particular to an infrared night vision image efficient communication method based on image compression.

Background

With the progress of the infrared night vision device, the frequency of the collected infrared night vision image is increased, more and more infrared night vision image data are obtained, and a large amount of infrared night vision image data need to be transmitted in order to facilitate real-time understanding of the specific conditions of the detection area of the infrared night vision device. In the process of image data transmission, an infrared night vision image is formed by illuminating a target by using an infrared searchlight and receiving reflection and infrared radiation, and the infrared night vision image has the defects of low contrast, fuzzy target edge detail information and the like, so that a method capable of efficiently and accurately transmitting image data is needed.

With the development of an image compression technology, in order to overcome the problem that the details of an infrared night vision image to be transmitted and compressed are not clear, bit layering is used for reconstructing the infrared night vision image, the traditional bit layering mainly selects a plurality of bit layering at random, and the random selection of the bit layering easily causes the loss of the detail information of the key position of the image after reconstruction processing, so that certain defect exists. The method considers the non-local self-similarity and local smooth characteristic of an image, improves a traditional full-variation model, only constructs a weight coefficient for a high-frequency component of the image, further compresses the reconstructed image, and protects the detail information of the image to a certain extent, but in the process of processing the image, the algorithm of the method inevitably has a structure without repeatability and image data damaged by noise, so that the texture detail of the transmitted compressed image is blurred, and the method has large calculation amount and long time consumption.

Disclosure of Invention

In order to solve the technical problem that the definition of detail information of the existing compressed image to be transmitted is poor, the invention aims to provide an infrared night vision image high-efficiency communication method based on image compression, and the adopted technical scheme is as follows:

an embodiment of the invention provides an infrared night vision image high-efficiency communication method based on image compression, which comprises the following steps:

acquiring N frames of infrared night vision images to be transmitted, and further determining target area images to be reconstructed corresponding to the M frames of target infrared night vision images;

acquiring each corner of a target area image to be reconstructed, and determining a correlation index corresponding to any two target area images to be reconstructed according to the gray value of each corner of any two target area images to be reconstructed;

determining each group of target area images to be reconstructed with the same target according to the correlation indexes corresponding to any two target area images to be reconstructed, and further determining the information content index corresponding to each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructed;

acquiring the number of connected domains in each bit layer image of a first target area image to be reconstructed, and determining a weighted value corresponding to each bit layer image of the first target area image to be reconstructed according to an information content index corresponding to each bit layer image of the first target area image to be reconstructed and the number of the connected domains thereof;

acquiring a time sequence interval and a corresponding information entropy between a first target area image to be reconstructed and a second target area image to be reconstructed in each group of target area images to be reconstructed, and determining a weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed according to the time sequence interval and the corresponding information entropy between the first target area image to be reconstructed and the second target area image to be reconstructed in each group of target area images to be reconstructed and information content indexes corresponding to each bit layer image of the first target area image to be reconstructed;

determining the weight value of each bit layer image corresponding to the second target area image to be reconstructed according to the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed and the weight value corresponding to each bit layer image of the first target area image to be reconstructed, and further determining the weight value of each bit layer image of each target area image to be reconstructed;

and according to the weight value of each bit layer image of each target area image to be reconstructed, obtaining each target area image to be reconstructed after reconstruction processing, and compressing and transmitting each target area image to be reconstructed after reconstruction processing.

Further, the step of determining the correlation indexes corresponding to any two target area images to be reconstructed includes:

determining a gray level mean value of all corner points of any two target area images to be reconstructed according to the gray level values of all corner points of any two target area images to be reconstructed;

and determining the correlation indexes corresponding to any two target area images to be reconstructed according to the gray values of all the corners of any two target area images to be reconstructed and the gray average value of all the corners.

calculating the difference value between the gray value of each corner of any two target area images to be reconstructed and the gray average value of all the corners corresponding to the gray value, taking the accumulated value obtained by multiplying the difference values as the numerator of the ratio, further calculating the corner gray value variance value corresponding to any two target area images to be reconstructed, taking the product obtained by multiplying the corner gray value variance value as the denominator of the ratio, wherein the ratio is the correlation index corresponding to any two target area images to be reconstructed.

Further, the step of determining the target area image to be reconstructed corresponding to the M frames of target infrared night vision images includes:

determining M frames of target infrared night vision images according to the N frames of infrared night vision images to be transmitted;

and inputting the M frames of target infrared night vision images into a pre-constructed and trained semantic segmentation network, and outputting target area images to be reconstructed corresponding to the M frames of target infrared night vision images.

Further, the step of determining each group of target region images to be reconstructed having the same target includes:

if the correlation indexes corresponding to any two target area images to be reconstructed are larger than the correlation threshold value, judging that the two target area images to be reconstructed belong to target area images to be reconstructed with the same target, otherwise, judging that the two target area images to be reconstructed do not belong to target area images to be reconstructed with the same target;

and dividing the target area images to be reconstructed with the same target into a group according to the comparison result of the correlation indexes corresponding to any two target area images to be reconstructed and the correlation threshold, so as to obtain each group of target area images to be reconstructed with the same target.

Further, the step of determining the information content index corresponding to each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructed includes:

constructing a sliding window with a preset size, and enabling the sliding window to slide on each bit layer image of a first target area image to be reconstructed in each group of target area images to be reconstructed according to a preset step length to obtain each sliding window area corresponding to each bit layer image;

determining the information content of each sliding window area corresponding to each bit layer image according to the pixel value of each pixel point in each sliding window area corresponding to each bit layer image;

the method comprises the steps of obtaining the number of sliding window areas of each bit layer image, and determining information content indexes corresponding to each bit layer image of a first target area image to be reconstructed in each group of target area images to be reconstructed according to the information content of each sliding window area corresponding to each bit layer image and the number of the sliding window areas.

Further, the step of determining the weight value corresponding to each bit layer image of the first target region image to be reconstructed includes:

and calculating the product of the information content index of any bit layer image of the first target region image to be reconstructed and the number of the connected domains, further constructing a logarithmic function of the product of the information content index and the number of the connected domains with a preset value as a base, and taking the value of the logarithmic function as the weight value of the bit layer image.

Further, the step of determining the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed includes:

determining information entropy difference values corresponding to a first target area image to be reconstructed and a second target area image to be reconstructed according to information entropies corresponding to the first target area image to be reconstructed and the second target area image to be reconstructed in each group of target area images to be reconstructed;

and calculating the product of the information entropy difference value corresponding to the first target area image to be reconstructed and the second target area image to be reconstructed, the time sequence interval between the first target area image to be reconstructed and the second target area image to be reconstructed and the information content index corresponding to any bit layer image of the first target area image to be reconstructed, taking the product as the weight compensation deviation value corresponding to the bit layer image of the second target area image to be reconstructed, and further determining the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed.

Further, the step of determining the weight value of each bit layer image corresponding to the second target region image to be reconstructed includes:

and adding the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed with the weight value corresponding to each bit layer image of the first target area image to be reconstructed, and taking the added value as the weight value of each bit layer image corresponding to the second target area image to be reconstructed.

Further, the step of obtaining the target area image to be reconstructed after the reconstruction processing includes:

selecting a preset number of bit layer images from the weight values corresponding to the bit layer images according to the weight values corresponding to the bit layer images of each target area image to be reconstructed;

and reconstructing each target area image to be reconstructed according to the front preset number of bit layer images corresponding to each target area image to be reconstructed, so as to obtain each target area image to be reconstructed after reconstruction processing.

The invention has the following beneficial effects:

the invention provides an efficient communication method of infrared night vision images based on image compression, which utilizes an image data processing technology to process and analyze N frames of acquired infrared night vision images to be transmitted, acquire M frames of target infrared night vision images, further obtain weighted values corresponding to bit layer images of all target area images to be reconstructed corresponding to the M frames of target infrared night vision images, realize reconstruction processing of all target area images to be reconstructed, ensure clearer detail content of the reconstructed images, help to ensure detail content of key parts of the infrared night vision images while ensuring high compression ratio, enable the transmitted image data to be clearer and more accurate, and finally compress and transmit all the reconstructed target area images after reconstruction processing. The method comprises the steps of obtaining a target area image to be reconstructed corresponding to M frames of target infrared night vision images, screening N frames of infrared night vision images to be transmitted, and eliminating background area detail information of the target infrared night vision images, wherein the information accuracy of the images is improved to a certain extent, and the calculated amount of image data is reduced; and determining the correlation indexes corresponding to any two target area images to be reconstructed according to the gray values of all corner points of any two target area images to be reconstructed. In order to judge whether the two images have the same target, an index value is determined according to the image characteristic information of the two images, and the accuracy of the judgment result can be effectively improved by utilizing the index value for judgment; in order to facilitate the subsequent analysis of the target area image to be reconstructed, each group of target area images to be reconstructed with the same target are determined, which is beneficial to determining the weight values corresponding to each bit layer image of each target area image to be reconstructed more accurately and rapidly in the subsequent process. In order to improve the image information details of a target area image of a target infrared night vision image, determining a weight value corresponding to each bit layer image of the target area image to be reconstructed, and determining the information content of each bit layer image of the target area image to be reconstructed in order to determine the weight value; and determining the weight value corresponding to each bit layer image of the first target area image to be reconstructed according to the information content index corresponding to each bit layer image of the first target area image to be reconstructed and the number of the connected domains thereof. The information content index and the number of the connected domains are positively correlated with the weight value, the larger the information content index and the number of the connected domains are, the larger the weight value is, the more accurate the weight value obtained by the two indexes is, and the calculation of the weight value corresponding to each bit layer image is beneficial to the accurate reconstruction of the subsequent target area image to be reconstructed; in order to facilitate the subsequent determination of the weight value of each bit layer image corresponding to the second target area image to be reconstructed, determining a weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed; determining the weight value of each bit layer image corresponding to the second target area image to be reconstructed according to the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed and the weight value corresponding to each bit layer image of the first target area image to be reconstructed, and further determining the weight value corresponding to each bit layer image of each target area image to be reconstructed. According to the weight corresponding to each bit layer image of the previous target area image to be reconstructed in each group of target area images to be reconstructed and the weight compensation deviation value corresponding to the weight, the weight corresponding to each bit layer image of the next target area image to be reconstructed is determined, so that the accuracy of the weight corresponding to each bit layer image is improved, the calculated amount is reduced, and the image processing speed efficiency and the image definition of transmitting the infrared night vision image are further ensured.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.

FIG. 1 is a flow chart of an infrared night vision image high-efficiency communication method based on image compression according to the invention.

Detailed Description

To further explain the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description of the embodiments, structures, features and effects of the technical solutions according to the present invention will be given with reference to the accompanying drawings and preferred embodiments. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

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 invention provides an infrared night vision image high-efficiency communication method based on image compression, as shown in figure 1, the method comprises the following steps:

(1) and acquiring N frames of infrared night vision images to be transmitted, and further determining target area images to be reconstructed corresponding to the M frames of target infrared night vision images.

The method includes the steps that N frames of infrared night vision images to be transmitted are obtained, the N frames of infrared night vision images to be transmitted in a certain period are collected in real time by an infrared camera of an infrared night vision instrument, the infrared night vision images are continuously distributed according to a time sequence, and the N frames of infrared night vision images can be used for image feature information needing to be analyzed later.

Determining a target area image to be reconstructed corresponding to the M frames of target infrared night vision images according to the N frames of infrared night vision images to be transmitted, wherein the method comprises the following steps:

and (1-1) determining M frames of target infrared night vision images according to the N frames of infrared night vision images to be transmitted.

In this embodiment, the N frames of infrared night vision images to be transmitted are input into a pre-established and trained preset target detection network, a determination result of whether the N frames of infrared night vision images to be transmitted have the preset target is output, the infrared night vision images with the preset target are obtained based on the determination result corresponding to the N frames of infrared night vision images to be transmitted, the infrared night vision images are called target infrared night vision images, and the preset target can be set by an implementer according to the shot infrared night vision images. In addition, the infrared night vision image without the preset target is subjected to high-compression-rate processing and transmission, and the infrared night vision image with the preset target and the infrared night vision image without the preset target are separately subjected to compression transmission, so that the image information loss of the infrared night vision image with the preset target under the condition of high compression rate can be avoided, and the compression rate of image compression is ensured. The training and construction processes of the preset target detection network are the prior art and are not within the protection scope of the invention, and the detailed description is not provided herein.

It should be noted that the number M of the target infrared night vision images may be equal to the number N of the infrared night vision images to be transmitted, which indicates that all the N frames of infrared night vision images to be transmitted have the preset target, and certainly, the number M of the target infrared night vision images may also be unequal to the number N of the infrared night vision images to be transmitted, which indicates that there is an infrared night vision image to be transmitted in the N frames of infrared night vision images to be transmitted, which does not have the preset target.

And (1-2) inputting the M frames of target infrared night vision images into a pre-constructed and trained semantic segmentation network, and outputting target area images to be reconstructed corresponding to the M frames of target infrared night vision images.

Based on the semantic segmentation network technology, the background part and the preset target part of the M-frame target infrared night vision image to be transmitted can be better identified and analyzed. The network structure of the semantic segmentation network is an Encoder-Decoder structure, the semantic segmentation network performs convolution operation through an Encoder to extract features, the output result of the Encoder is a feature map, and the feature map is operated through a Decoder to obtain a target area image. In this embodiment, the input image of the semantic segmentation network is an M-frame target infrared night vision image, and the output image is a target area image to be reconstructed corresponding to the M-frame target infrared night vision image, which specifically includes: inputting the M frames of target infrared night vision images into a pre-constructed and trained semantic segmentation network, marking the pixel points of the preset target area of the M frames of target infrared night vision images as 1, marking the pixel points of other areas as 0, wherein the preset target area of the target infrared night vision images can be a plurality of areas or a single area, the number of the specific preset target areas can be determined according to the actual condition of the target infrared night vision images, and outputting the target area images to be reconstructed corresponding to the M frames of target infrared night vision images. The implementation process of the semantic segmentation network is the prior art and is not within the protection scope of the present invention, and is not described in detail herein.

It should be noted that semantic analysis is performed on the M-frame target infrared night vision image to obtain a highlight portion in the M-frame target infrared night vision image, the highlight portion may be a human or an animal, a preset target may be a human, an animal or another object, the highlight portion in the image is marked to facilitate subsequent processing and analysis of image feature data of a highlight portion area image, and the highlight portion area image may be referred to as a target area image to be reconstructed.

(2) Acquiring each corner of a target area image to be reconstructed, and determining a correlation index corresponding to any two target area images to be reconstructed according to the gray value of each corner of any two target area images to be reconstructed.

Each corner of the target area image to be reconstructed is obtained in order to determine whether the preset target of each target area image to be reconstructed is the same. And performing corner detection on the target area image to be reconstructed corresponding to the M frames of target infrared night vision images to obtain each corner of the target area image to be reconstructed corresponding to the M frames of target infrared night vision images. The process of detecting the corner points of the image is the prior art and is not within the scope of the present invention, and will not be elaborated herein.

Determining a correlation index corresponding to any two target area images to be reconstructed according to the gray value of each corner of any two target area images to be reconstructed, wherein the method comprises the following steps:

and (2-1) calculating the gray average value of all corner points in any two target area images to be reconstructed according to the gray values of all corner points in any two target area images to be reconstructed.

In order to facilitate subsequent calculation of correlation indexes corresponding to any two target area images to be reconstructed, based on gray values of all corner points in any two target area images to be reconstructed, a gray average value of all corner points in any two target area images to be reconstructed is calculated by using correlation knowledge of mathematical modeling, and the calculation formula is as follows:

wherein the content of the first and second substances,

the gray level mean value of all corner points in any two target area images to be reconstructed,Kfor the number of angular points in any two target region images to be reconstructed, c _k For the first of any two target region images to be reconstructedkThe grey value of each corner point.

It should be noted that the gray level mean value of all the corner points in any two target region images to be reconstructed may be referred to as the corner point gray level mean value corresponding to any two target region images to be reconstructed, and each target region image to be reconstructed has the corner point gray level mean value corresponding thereto.

And (2-2) determining the correlation indexes corresponding to any two target area images to be reconstructed according to the gray values of all the corners of any two target area images to be reconstructed and the gray average value of all the corners.

It should be noted that the correlation index corresponding to the two target region images to be reconstructed may be understood as the similarity between the key points of the two target region images to be reconstructed, and the greater the similarity between the key points of the two target region images to be reconstructed, the higher the correlation index corresponding to the two target region images to be reconstructed is. In this embodiment, the key points of the target region image to be reconstructed are corner points, so that the similarity between the key points of the two target region images to be reconstructed can be determined based on the gray values of the corner points of any two target region images to be reconstructed and the gray average value of all the corner points, and the determined correlation index has a certain referential property.

In this embodiment, knowing the gray value of each corner point of any two target area images to be reconstructed and the gray average value of all corner points, calculating the difference between the gray value of each corner point of any two target area images to be reconstructed and the gray average value of all corner points corresponding to the gray value, taking the accumulated value obtained by multiplying the difference as the numerator of a ratio, further calculating the gray variance value of the corner point corresponding to any two target area images to be reconstructed, taking the product obtained by multiplying the gray variance value of the corner point as the denominator of the ratio, where the ratio is a correlation index corresponding to any two target area images to be reconstructed. The process of calculating the angular point gray scale variance value is an existing algorithm, and will not be described herein too much. According to the gray value of each corner of any two target area images to be reconstructed and the gray average value of all corners, the calculation formula of the correlation index can be as follows based on the mathematical modeling correlation knowledge:

wherein the content of the first and second substances,Cfor the correlation indexes corresponding to any two target area images to be reconstructed,

for the first of any two target area images to be reconstructediThe gray value of an individual corner point,

the gray level mean value of all corner points in one target area image to be reconstructed in any two target area images to be reconstructed,

for the first of any two target region images to be reconstructedjThe gray value of an individual corner point,

the gray level mean value of all corner points in the other target area image to be reconstructed of any two target area images to be reconstructed,Ifor the number of corner points in one of any two target region images to be reconstructed,Jthe number of the corner points in the other to-be-reconstructed target area image of any two to-be-reconstructed target area images.

It should be noted that the denominator of the ratio in the calculation formula is the angular point gray variance, the smaller the angular point gray variance is, the higher the correlation between the two target area images to be reconstructed is, the numerator of the ratio is the accumulated value obtained by multiplying the angular point gray difference, and the larger the accumulated value is, the higher the correlation between the two target area images to be reconstructed is. The larger the correlation index corresponding to any two target area images to be reconstructed is, the more likely the preset targets in the two target area images to be reconstructed are to be the same.

(3) Determining each group of target area images to be reconstructed with the same target according to the correlation indexes corresponding to any two target area images to be reconstructed, and further determining the information content index corresponding to each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructed, wherein the method comprises the following steps of:

(3-1) determining each group of target area images to be reconstructed with the same target according to the correlation indexes corresponding to any two target area images to be reconstructed, wherein the steps comprise:

(3-1-2) if the correlation indexes corresponding to any two target area images to be reconstructed are larger than a correlation threshold, determining that the two target area images to be reconstructed belong to target area images to be reconstructed with the same target, otherwise, determining that the two target area images to be reconstructed do not belong to target area images to be reconstructed with the same target.

(3-1-3) dividing the target area images to be reconstructed with the same target into a group according to the comparison result of the correlation indexes corresponding to any two target area images to be reconstructed and the correlation threshold, thereby obtaining each group of target area images to be reconstructed with the same target.

In this embodiment, the correlation indexes corresponding to two consecutive target area images to be reconstructed, which are both greater than the correlation threshold value 0.6, are divided into one group, so as to obtain each group of target area images to be reconstructed having the same target, where the preset targets of each target area image to be reconstructed in each group of target area images to be reconstructed are the same.

Assuming that the number of the target area images to be reconstructed is 5, the correlation index corresponding to the first target area image to be reconstructed and the second target area image to be reconstructed is greater than the correlation threshold 0.6, the correlation index corresponding to the second target area image to be reconstructed and the third target area image to be reconstructed is also greater than the correlation threshold 0.6, and the correlation index corresponding to the third target area image to be reconstructed and the fourth target area image to be reconstructed is less than the correlation threshold 0.6, it is indicated that the preset targets of the first target area image to be reconstructed, the second target area image to be reconstructed and the third target area image to be reconstructed are the same, and the three target area images to be reconstructed are the first group of target area images to be reconstructed with the same target. And if the correlation index corresponding to the fourth target area image to be reconstructed and the fifth target area image to be reconstructed is greater than the correlation threshold value of 0.6, it indicates that the preset targets of the fourth target area image to be reconstructed and the fifth target area image to be reconstructed are the same, and the fourth target area image to be reconstructed and the fifth target area image to be reconstructed are a second group of target area images to be reconstructed with the same targets. Therefore, the division of the target area image to be reconstructed is completed, and compared with the existing artificial visual grouping, the grouping based on the relevance indexes is more accurate.

And (3-2) determining information content indexes corresponding to bit layer images of a first target area image to be reconstructed in each group of target area images to be reconstructed according to each group of target area images to be reconstructed with the same target.

Obtaining each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructed, wherein the gray value of each pixel point in the common 256-level gray image is composed of 8 bits, and separating the 8 bits respectively to form 8 new images, wherein the 8 new images are called as bit layer images. And acquiring each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructed based on the first target area image to be reconstructed in each group of target area images to be reconstructed with the same target. The process of obtaining a bit-layer image of an image is prior art and is not within the scope of the present invention, and will not be described in detail herein.

Determining information content indexes corresponding to bit layer images of a first target area image to be reconstructed in each group of target area images to be reconstructed according to pixel values of each pixel point of each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructed, wherein the steps comprise:

(3-2-1) constructing a sliding window with a preset size, and enabling the sliding window to slide on each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructed according to a preset step length to obtain each sliding window area corresponding to each bit layer image.

It should be noted that, in this embodiment, the preset size of the sliding window is 3 × 3, the preset step size is 1, and both the preset size and the preset step size of the sliding window can be set by an implementer according to the specific size of the image. The process of constructing the sliding window is prior art and is not within the scope of the present invention, and will not be described in detail herein.

And (3-2-2) determining the information content of each sliding window area corresponding to each bit layer image according to the pixel value of each pixel point in each sliding window area corresponding to each bit layer image of the first target area image to be reconstructed.

In this embodiment, the image information of the bit layer image only contains 0 and 1, that is, the pixel value of each pixel in the bit layer image is 0 or 1, the larger the number of pixels with pixel values of 1 indicates the larger the information content, the more the number of pixels with pixel values is, the statistics is performed on the number of pixels with pixel values of 0 and pixel values of 1 in each sliding window region corresponding to each bit layer image of the first target region image, and the information content of each sliding window region corresponding to each bit layer image is determined by using the relevant knowledge of mathematical modeling, and the calculation formula is as follows:

wherein the content of the first and second substances,pfor the information content of each sliding window region corresponding to each bit layer image of the first target region image to be reconstructed,

the number of pixel points with the pixel value of 0 in the information content of each sliding window area corresponding to each bit layer image,

the number of pixel points with the information content pixel value of 1 of each sliding window area corresponding to each bit layer image of the first target area image to be reconstructed is determined,

and

the information content coefficients of the sliding window regions corresponding to the bit layer images of the first target region image to be reconstructed can be set by an implementer according to specific conditions.

It should be noted that, the larger the number of pixels with a pixel value of 1 in a certain sliding window region is, the larger the information content corresponding to the sliding window region will be, and the larger the number of pixels with a pixel value of 0 in a certain sliding window region is, the smaller the information content corresponding to the sliding window region will be.

(3-2-3) acquiring the number of sliding window areas of each bit layer image of the first target area image to be reconstructed, and determining information content indexes corresponding to each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructed according to the information content of each sliding window area corresponding to each bit layer image and the number of the sliding window areas.

In this embodiment, in order to subsequently determine the information content index corresponding to the bit layer image, the number of sliding window regions of each bit layer image of the first target region image to be reconstructed is counted, and the number of sliding window regions is one of the important indexes for determining the information content index, so that the number of sliding window regions of each bit layer image of the first target region image to be reconstructed is obtained at this time.

In order to comprehensively analyze the information content index corresponding to each bit layer image of a first target area image to be reconstructed in each group of target area images to be reconstructed, the information content of each sliding window area corresponding to each bit layer image is accumulated according to the number of the sliding window areas of each bit layer image of the first target area image to be reconstructed and the information content of each sliding window area corresponding to each bit layer image, the accumulated value is used as the information content index corresponding to the corresponding bit layer image, and the calculation formula is as follows:

wherein the content of the first and second substances,Pfor the information content index corresponding to each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructed,p _l for the second corresponding to each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructedlThe information content of the individual sliding window regions,Land the number of the sliding window areas of each bit layer image of the first target area image to be reconstructed in each group of target area images to be reconstructed.

According to the calculation formula of the information content index, the larger the accumulated value of the information content of each sliding window area corresponding to any bit layer image is, the larger the information content index corresponding to the bit layer image is, that is, the greater the importance degree of the bit layer image in each bit layer is.

(4) The method comprises the steps of obtaining the number of connected domains in each bit layer image of a first target area image to be reconstructed, and determining the weight value corresponding to each bit layer image of the first target area image to be reconstructed according to the information content index corresponding to each bit layer image of the first target area image to be reconstructed and the number of the connected domains.

It should be noted that, the more the number of connected domains in the bit layer image, the more the number of preset targets in the bit layer image, the higher the attention degree of the bit layer image should be, the more the information content of the bit layer image is, the more important the bit layer image is in the process of image reconstruction, so the number of connected domains in each bit layer image of the first target area image to be reconstructed is counted.

In this embodiment, based on the information content index corresponding to each bit layer image of the first target area image to be reconstructed obtained in step (3-2-3) and the number of connected domains thereof, a product of the information content index of any one bit layer image of the first target area image to be reconstructed and the number of connected domains is calculated, a logarithmic function of the product of the information content index and the number of connected domains based on a preset value is further constructed, a value of the logarithmic function is used as a weighted value of the bit layer image, a weighted value corresponding to each bit layer image of the first target area image to be reconstructed is obtained according to the same process of determining the weighted value of the bit layer image, and a calculation formula of the weighted value corresponding to each bit layer image is known by using relevant knowledge of mathematical modeling:

wherein the content of the first and second substances,

the weight values corresponding to the bit layer images of the first target area image to be reconstructed,

the number of connected domains in each bit layer image of the first target area image to be reconstructed,Plog () is a logarithm function, 10 is a preset value of this embodiment, and an implementer can set itself according to specific practical situations.

In order to facilitate subsequent data analysis on the weight values corresponding to the bit layer images, the embodiment normalizes the weight values corresponding to the bit layer images of the first target area image to be reconstructed to obtain the weight values corresponding to the bit layer images after the normalization. The number and the information content index of the connected domains in the bit layer image are positively correlated with the weight value corresponding to the bit layer image, and the larger the number and the information content index of the connected domains in the bit layer image are, the larger the weight value corresponding to the bit layer image is.

(5) The method comprises the steps of obtaining a time sequence interval between a first target area image to be reconstructed and a second target area image to be reconstructed in each group of target area images to be reconstructed and corresponding information entropy, and determining a weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed according to the time sequence interval between the first target area image to be reconstructed and the second target area image to be reconstructed in each group of target area images to be reconstructed and the corresponding information entropy and information content indexes corresponding to each bit layer image of the first target area image to be reconstructed.

First, it should be noted that, in this embodiment, compressed image transmission is performed based on M frames of target infrared night vision images, each bit layer image corresponding to each frame of target area image to be reconstructed corresponding to each frame of target infrared night vision image is provided, and if each bit layer image of each target area image to be reconstructed corresponding to each frame of target infrared night vision image is determined according to the step of the weight value corresponding to each bit layer image of the first target area image to be reconstructed, time consumed by the step is too long. In addition, since the transmission of the compressed image is real-time, if the weight values of the bit layer images of the target area image to be reconstructed are calculated one by one and then calculated, the real-time performance is poor, and the transmission speed of the compressed data is greatly reduced. In order to ensure the transmission speed and real-time performance of the compressed image, the weight values of the bit layer images of the second target area image to be reconstructed are calculated according to the weight values of the bit layer images of the first target area image to be reconstructed in each group of target area images to be reconstructed, namely the weight values of the bit layer images of the next target area image to be reconstructed are calculated according to the weight values of the bit layer images of the previous target area image to be reconstructed. When calculating the weight value corresponding to each bit layer image of each target area image to be reconstructed, firstly, the weight compensation deviation value corresponding to each bit layer image of a second target area image to be reconstructed needs to be determined, and the steps include:

and (5-1) acquiring a time sequence interval between a first target area image to be reconstructed and a second target area image to be reconstructed in each group of target area images to be reconstructed and corresponding information entropy.

In order to facilitate the subsequent determination of the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed, the time sequence interval between the first target area image to be reconstructed and the second target area image to be reconstructed in each group of target area images to be reconstructed and the corresponding information entropy are obtained. The time sequence interval represents the time difference of two target area images to be reconstructed, the larger the time difference is, that is, the longer the time interval is, the larger the influence on the weight compensation deviation value is, the larger the value to be subjected to weight compensation is, and the information entropy represents the degree of the overall change of the image characteristic information of the two target area images to be reconstructed, and the larger the information entropy difference corresponding to the two target area images to be reconstructed is, the larger the value to be subjected to weight compensation is.

(5-2) determining a weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed according to a time sequence interval between the first target area image to be reconstructed and the second target area image to be reconstructed in each group of target area images to be reconstructed, corresponding information entropy and information content indexes corresponding to each bit layer image of the first target area image to be reconstructed, wherein the steps comprise:

(5-2-1) determining information entropy difference values corresponding to the first target area image to be reconstructed and the second target area image to be reconstructed according to the information entropy corresponding to the first target area image to be reconstructed and the second target area image to be reconstructed in each group of target area images to be reconstructed.

In this embodiment, in order to determine the difference degree between the image characteristics of the first target area image to be reconstructed and the second target area image to be reconstructed in each set of target area images to be reconstructed, the information entropy of the first target area image to be reconstructed in each set of target area images to be reconstructed is subtracted from the information entropy of the second target area image to be reconstructed to obtain the information entropy difference value corresponding to the first target area image to be reconstructed and the second target area image, and the calculation formula is as follows:

wherein the content of the first and second substances,

for the information entropy difference value corresponding to the first target area image to be reconstructed and the second target area image to be reconstructed in each group of target area images to be reconstructed,E1 is the information entropy of the first target area image to be reconstructed in each group of target area images to be reconstructed,Eand 2 is the information entropy of the second target area image to be reconstructed in each group of target area images to be reconstructed.

It should be noted that the information entropy difference corresponding to the first target area image to be reconstructed and the second target area image to be reconstructed may be used to subsequently determine the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed. The weight compensation deviation value is determined based on the information entropy difference value, so that the accuracy of the subsequently determined weight compensation deviation value can be improved.

(5-2-2) calculating the product of the information entropy difference value corresponding to the first target area image to be reconstructed and the second target area image to be reconstructed, the time sequence interval between the first target area image to be reconstructed and the second target area image to be reconstructed and the information content index corresponding to any bit layer image of the first target area image to be reconstructed, taking the product as the weight compensation deviation value corresponding to the bit layer image of the second target area image to be reconstructed, and further determining the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed.

In this embodiment, the weight compensation deviation values corresponding to the bit layer images of the second target area image to be reconstructed, the information entropy difference values corresponding to the first target area image to be reconstructed and the second target area image to be reconstructed, the time sequence interval between the first target area image to be reconstructed and the second target area image to be reconstructed, the information content index corresponding to each bit layer image of the first target area image to be reconstructed and the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed are all positively correlated from 3 aspects, the information entropy difference value and the time sequence interval are coefficients, the information content index corresponding to any bit layer image of the first target area image to be reconstructed is the key for calculating the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed. Taking the calculation of the weight compensation deviation value corresponding to any bit layer image of the second target area image to be reconstructed as an example, the calculation formula of the weight compensation deviation value is as follows:

wherein the content of the first and second substances,

compensating the deviation value for the weight corresponding to any bit layer image of the second target area image to be reconstructed,

the information entropy difference corresponding to the first target area image to be reconstructed and the second target area image to be reconstructed,

for the temporal interval between the first target region image to be reconstructed and the second target region image to be reconstructed,

and the index is the information content index corresponding to the bit layer image of the first target area image to be reconstructed.

It should be noted that each bit layer image of the first target region image to be reconstructed is compared with each bit layer image of the second target region image to be reconstructed, that is, the number of bit layer images of each target region image to be reconstructed is the same. The calculation formula of the weight compensation deviation value can show that the larger the information entropy difference, the larger the time sequence interval and the larger the information content index, the larger the weight compensation deviation value is. And determining the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed by referring to the determination process of the weight compensation deviation value corresponding to any bit layer image of the second target area image to be reconstructed.

(6) Determining the weight value of each bit layer image corresponding to the second target area image to be reconstructed according to the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed and the weight value corresponding to each bit layer image of the first target area image to be reconstructed, and further determining the weight value corresponding to each bit layer image of each target area image to be reconstructed.

In this embodiment, in order to perform reconstruction processing on the second target region image to be reconstructed, it is necessary to determine a weight value corresponding to each bit layer image of each target region image to be reconstructed. The method specifically comprises the following steps: and adding the weight compensation deviation value corresponding to each bit layer image of the second target area image to be reconstructed with the weight value corresponding to each bit layer image of the first target area image to be reconstructed, and taking the added value as the weight value of each bit layer image corresponding to the second target area image to be reconstructed. Each bit layer image corresponding to the second target area image to be reconstructed has a weight value corresponding to the second target area image, and the calculation formula is as follows:

wherein the content of the first and second substances,

the weight value of each bit layer image corresponding to the second target area image to be reconstructed,

for each bit of the first target region image to be reconstructedThe weight value corresponding to the layer image,

and compensating the deviation value for the weight corresponding to each bit layer image of the second target area image to be reconstructed.

Based on the weight values of the bit layer images corresponding to the second target area image to be reconstructed in each group of target area images to be reconstructed and the image characteristics of the third target area image to be reconstructed, referring to the determination process of the weight values of the bit layer images corresponding to the second target area image to be reconstructed in the steps (4) to (6), obtaining the weight values of the bit layer images corresponding to the third target area image to be reconstructed, namely calculating the weight values of the bit layer images corresponding to the next target area image to be reconstructed based on the weight values of the bit layer images corresponding to the previous target area image to be reconstructed in each group of target area images to be reconstructed and the image characteristics of the next target area image to be reconstructed, further obtaining the weight values corresponding to the bit layer images of the target area images to be reconstructed, which effectively improves the image calculation time, the efficiency of compressed image transmission and communication is improved.

(7) And obtaining each target area image to be reconstructed after reconstruction processing according to the weight value corresponding to each bit layer image of each target area image to be reconstructed, and compressing and transmitting each target area image to be reconstructed after reconstruction processing.

(7-1) obtaining each target area image to be reconstructed after reconstruction processing according to the weight value corresponding to each bit layer image of each target area image to be reconstructed, wherein the steps comprise:

(7-1-1) selecting a preset number of bit layer images from the weight values corresponding to the bit layer images according to the weight values corresponding to the bit layer images of the target area image to be reconstructed.

In this embodiment, the weighted values corresponding to the bit layer images of each target area image to be reconstructed are sorted according to a certain sequence, a first preset number of bit layer images with a larger weighted value in the sorting sequence are selected, the preset number is set to 3, that is, the first 3 bit layer images with a larger weighted value are selected from the bit layer images of each target area image to be reconstructed, the preset number is determined according to prior knowledge, and an implementer can set the preset number according to the complexity of the image to be reconstructed.

It should be noted that, when the target area image to be reconstructed is reconstructed, a more important bit layer of the target area image to be reconstructed is generally selected to reconstruct the target area image to be reconstructed, so as to reduce the amount of data that needs to be transmitted in the reverse direction.

And (7-1-2) reconstructing each target area image to be reconstructed according to the front preset number of bit layer images corresponding to each target area image to be reconstructed, thereby obtaining each target area image to be reconstructed after reconstruction processing.

In this embodiment, based on the first 3 bit layer images with a larger weight value corresponding to each target area image to be reconstructed, the target area images to be reconstructed corresponding to the 3 bit layer images with a larger weight value are reconstructed to obtain each target area image to be reconstructed after the reconstruction processing, image details of each target area image to be reconstructed after the reconstruction processing are clearer, and image information of the transmitted compressed image is more accurate. The process of reconstructing the image is prior art and is not within the scope of the present invention, and will not be described in detail herein.

And (7-2) compressing and transmitting each target area image to be reconstructed after reconstruction processing according to each target area image to be reconstructed after reconstruction processing.

In this embodiment, since the compressed and transmitted image is an infrared image, and the pixel points in the infrared image are mainly formed according to the infrared radiation values emitted by the image, the redundancy of the pixel points in each reconstructed target area image after reconstruction processing is high, each reconstructed target area image after reconstruction processing is compressed by using a huffman coding compression method, and each reconstructed target area image after compression processing is transmitted to the terminal. So far, the embodiment realizes the compression transmission of the N frames of infrared night vision images to be transmitted. The compression implementation of huffman coding is prior art and is not within the scope of the present invention, and will not be elaborated here.

The above-mentioned embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; the modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present application, and are included in the protection scope of the present application.

Claims

1. An efficient infrared night vision image communication method based on image compression is characterized by comprising the following steps:

acquiring each corner of the target area image to be reconstructed, and determining a correlation index corresponding to any two target area images to be reconstructed according to the gray value of each corner of any two target area images to be reconstructed;

2. The method as claimed in claim 1, wherein the step of determining the correlation index corresponding to any two target area images to be reconstructed comprises:

determining a gray level mean value of all corner points of any two target area images to be reconstructed according to gray levels of all corner points of any two target area images to be reconstructed;

3. The method as claimed in claim 2, wherein the step of determining the correlation index corresponding to any two target area images to be reconstructed comprises:

4. The efficient communication method for infrared night vision images based on image compression as claimed in claim 1, wherein the step of determining the target area image to be reconstructed corresponding to the M frames of target infrared night vision images further comprises:

determining M frames of target infrared night vision images according to N frames of infrared night vision images to be transmitted;

5. The method as claimed in claim 1, wherein the step of determining each set of target region images to be reconstructed having the same target comprises:

6. The method as claimed in claim 1, wherein the step of determining the information content index corresponding to each bit layer image of the first target area image to be reconstructed in each set of target area images to be reconstructed further comprises:

and acquiring the number of sliding window areas of each bit layer image, and determining an information content index corresponding to each bit layer image of a first target area image to be reconstructed in each group of target area images to be reconstructed according to the information content of each sliding window area corresponding to each bit layer image and the number of the sliding window areas.

7. The infrared night vision image efficient communication method based on image compression as claimed in claim 1, wherein the step of determining the weight value corresponding to each bit layer image of the first target area image to be reconstructed comprises:

8. The method as claimed in claim 1, wherein the step of determining the weight compensation bias values corresponding to the bit layer images of the second target region image to be reconstructed comprises:

9. The method as claimed in claim 1, wherein the step of determining the weight value of each bit layer image corresponding to the second target area image to be reconstructed comprises:

10. The method as claimed in claim 1, wherein the step of obtaining the reconstructed target area images includes: