Panoramic photoelectric radar imaging brightness consistency processing method

Abstract

The invention relates to the technical field of image processing, and discloses a panoramic photoelectric radar imaging brightness consistency processing method. According to the method, when a system runs, each frame of image is collected, low-pass filtering processing is carried out before splicing, a low-pass filtering algorithm is adopted to obtain a low-frequency image Img (i), summation is carried out on continuous 50 frames of panoramic low-frequency images, then an average value is calculated, normalization processing is carried out on an average value image, the average value of low-frequency components is reduced to zero, and a corrected background image is obtained. According to the method, each frame of image of the panoramic photoelectric radar is corrected by using the correction background image, the corrected images are spliced and displayed, and the background image with non-uniform illumination is calculated by carrying out image subtraction processing on a real-time video and using the non-repeated imaging characteristic of a panoramic field image, so that the illumination is uniform, and the continuity of the overall effect of the image is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method for processing brightness consistency of panoramic photoelectric radar imaging. Background Art

[0002] When the panoramic photoelectric radar rotates at high speed, such as 1rps per second, the built-in infrared camera directly collects video images, which will cause serious ghosting and cannot form images normally. Therefore, the fast reflex mirror control technology is used for compensation. The fast reflex mirror control technology can make the panoramic photoelectric radar rotate at high speed. During the exposure time of each frame of image collected by the built-in infrared camera, the scene corresponding to each pixel remains unchanged, so that each frame can be clearly imaged. Each frame is spliced ​​together during the continuous 360° field of view to achieve the purpose of panoramic imaging.

[0003] Before the infrared light enters the detector, it passes through the galvanometer compensation in the optical path, which will cause energy inconsistency, making the illumination uneven and the overall effect of the 360° image observed by the human eye discontinuous. In order to solve this problem and ensure that it does not affect the real-time update of the video image, a simple and effective method is needed to deal with it. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the shortcomings of the prior art, the present invention provides a method for processing the brightness consistency of a panoramic photoelectric radar imaging. The method has the advantages of: when the system is running, each frame of the image is collected, low-pass filtering is performed before splicing, a low-pass filtering algorithm is used to obtain a low-frequency image Img(i), 50 consecutive frames of low-frequency images of the panoramic view are summed, the average value is calculated, the average image is normalized, the average value of the low-frequency component is normalized to 0, a corrected background image is obtained, each frame of the panoramic photoelectric radar is corrected by using the corrected background image, the corrected images are spliced ​​and displayed, image subtraction is performed on the real-time video, and the non-repetitive imaging features of the images of the panoramic field of view are used to calculate the uneven illumination background image, so that the illumination is uniform, the continuity of the overall image effect is improved, and the above-mentioned problems are solved.

[0006] (II) Technical solution

[0007] To achieve the above object, the present invention provides the following technical solution: a method for processing brightness consistency of panoramic photoelectric radar imaging, comprising the following steps:

[0008] S1. When the system is running, a medium-wave cooled infrared camera is used to collect each frame of the image, and a low-pass filtering process is performed before stitching. A low-pass filtering algorithm is used to obtain a low-frequency image Img(i);

[0009] S2, summing up 50 consecutive frames of low-frequency images of the panoramic view and then calculating the average value;

[0010] S3, normalizing the mean image, returning the average value of the low-frequency component to 0, and obtaining a corrected background image;

[0011] S4, using the corrected background image to correct each frame of the panoramic photoelectric radar;

[0012] S5. Stitching and displaying the corrected images.

[0013] Preferably, the S1 is subjected to low-pass filtering before stitching, and a low-pass filtering algorithm is used to obtain a formula for a low-frequency image Img(i) as follows:

[0014] ImgLF(i)=Filter2(Img(i))

[0015] In the formula, ImgLF(i) represents the low-frequency image of the i-th frame after low-pass filtering, Img(i) represents the image collected by the i-th frame, and Filter2 represents the operation of applying a two-dimensional filter to the input image.

[0016] Preferably, the formula for summing 50 consecutive frames of low-frequency images of the panoramic view in S2 is as follows:

[0017]

[0018] In the formula, ImgSum represents the sum of 50 frames of low-frequency images, ImgLF(i) represents the low-frequency image of the i-th frame after low-pass filtering, and i represents the number of image frames. The summation symbol indicates that the low-frequency images from the 1st frame to the 50th frame are summed.

[0019] Preferably, the formula for calculating the average value of 50 consecutive frames of low-frequency images of the panoramic view in S2 is as follows:

[0020] ImgMean=ImgSum / 50

[0021] In the formula, ImgMean represents the average value of 50 consecutive frames of low-frequency images of the periscopic view, and ImgSum represents the sum of 50 frames of low-frequency images.

[0022] Preferably, the formula for calculating the average value of the low-frequency component in S3 is as follows:

[0023]

[0024] In the formula, It represents the average value of the low-frequency component, M and N represent the height and width of the image, M*N represents the total number of pixels in the image, ImgMean(x, y) represents the pixel value at the coordinate (x, y), and the summation symbol represents the summation of each pixel in the image, and all pixel values ​​will be accumulated.

[0025] Preferably, S3 normalizes the mean image, returns the average value of the low-frequency component to 0, and obtains the formula for correcting the background image as follows:

[0026]

[0027] In the formula, ImgBase represents the corrected background image, ImgMean represents the average value of 50 consecutive low-frequency images of the surrounding view, Represents the average value of low-frequency components.

[0028] Preferably, when S4 uses the corrected background image to correct each frame of the panoramic photoelectric radar, histogram equalization and contrast enhancement are combined to enhance the image features. The correction process uses fast Fourier transform to accelerate the image processing speed.

[0029] Preferably, the formula for correcting each frame of the panoramic photoelectric radar by using the corrected background image in S4 is as follows:

[0030] ImgDst=ImgSrc-ImgBase

[0031] In the formula, ImgDst represents the image obtained after correction processing, ImgSrc represents the original frame image, which refers to the current frame image obtained from the panoramic photoelectric radar, and ImgBase represents the corrected background map.

[0032] Preferably, the formula for stitching the corrected images in S5 is as follows:

[0033]

[0034] In the formula, WarImg(x', y') represents the corresponding points in the spliced ​​image, (x, y) represents the points in the original image, H represents the homography matrix, and affine transformation is performed on each ImgDst.

[0035] Preferably, the formula for stitching and fusing the corrected images in S5 is as follows:

[0036] ImgResult(x, y)=α*Img1(x, y)+(1-α)*Img2(x, y)

[0037] In the formula, ImgResult(x, y) represents the fused image at position (x, y), that is, the final output image, Img1(x, y) represents the pixel value of the first input image, Img2(x, y) represents the pixel value of the second input image, and α represents the weight coefficient, which ranges from [0-1] and is used to control the proportion of the first image Img1 in the fusion.

[0038] Compared with the prior art, the present invention provides a method for processing brightness consistency of panoramic photoelectric radar imaging, which has the following beneficial effects:

[0039] The present invention collects each frame of image when the system is running, performs low-pass filtering before splicing, adopts a low-pass filtering algorithm to obtain a low-frequency image Img(i), sums 50 consecutive frames of low-frequency images of the surrounding view, calculates the average value, normalizes the average image, returns the average value of the low-frequency component to 0, obtains a corrected background image, uses the corrected background image to correct each frame of the surrounding photoelectric radar, splices and displays the corrected images, performs image subtraction processing on real-time video, uses the non-repetitive imaging characteristics of the surrounding field of view images, calculates the uneven illumination background image, makes the illumination uniform, and improves the continuity of the overall image effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the steps of the method of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] In order to solve the problem that the infrared light path will cause energy inconsistency when passing through the galvanometer compensation in the light path before entering the detector, resulting in uneven illumination and discontinuous overall image effect, a method for processing the brightness consistency of panoramic photoelectric radar imaging is proposed. Figure 1 , the method comprises the following steps:

[0043] S1. When the system is running, a medium-wave cooled infrared camera is used to collect each frame of the image, and a low-pass filtering process is performed before stitching. A low-pass filtering algorithm is used to obtain a low-frequency image Img(i);

[0044] ImgLF(i)=Filter2(Img(i))

[0045] Retaining important features Low-frequency images can usually retain most of the important information and contours in the image, while the high-frequency part mainly contains details and noise. For tasks such as target recognition and tracking, maintaining the main features of the image and discarding unnecessary details helps to improve the accuracy of subsequent analysis and processing. In the formula, ImgLF(i) represents the low-frequency image of the i-th frame after low-pass filtering, Img(i) represents the image collected by the i-th frame, and Filter2 represents the operation of applying a two-dimensional filter to the input image. The low-frequency image can better eliminate the influence of illumination changes during processing and enhance the brightness consistency in the image. This is especially important for panoramic optoelectronic radar imaging, because the uneven illumination under different scenes and measurement conditions may lead to large differences in imaging results;

[0046] S2, summing up 50 consecutive frames of low-frequency images of the panoramic view and then calculating the average value;

[0047] The summation formula is as follows:

[0048]

[0049] The summation process can effectively reduce the impact of random noise. Since noise mostly appears randomly, the summation of multiple frames can make the impact of noise in different frames offset each other, thereby improving the overall clarity of the image. In the formula, ImgSum represents the sum of 50 frames of low-frequency images, ImgLF(i) represents the low-frequency image of the i-th frame after low-pass filtering, and i represents the number of image frames. The summation symbol indicates the summation of low-frequency images from the 1st frame to the 50th frame. In dynamic scenes, by summing the frames, the instantaneous image changes caused by object movement or camera shake can be reduced, the stability of the image can be enhanced, and the subsequent analysis can be more reliable;

[0050] The formula for calculating the average value is as follows:

[0051] ImgMean=ImgSum / 50

[0052] Compared with single-frame images, the use of mean images can show clearer and more visually appealing image results, making the image processing effect more intuitive. In the formula, ImgMean represents the average value of 50 consecutive low-frequency images of periscopic vision, and ImgSum represents the sum of 50 low-frequency images. After calculating the average value, the retained features are mainly overall trends and important information, which are relatively less affected by accidental factors, and have better support for the accuracy of tasks such as target recognition.

[0053] S3, normalizing the mean image, returning the average value of the low-frequency component to 0, and obtaining a corrected background image;

[0054] The formula for calculating the average value of the low-frequency component is as follows:

[0055]

[0056] The low-frequency components usually contain the contour and global structure of the image. Calculating the average value can focus and highlight these important visual features. In the formula Represents the average value of low-frequency components, M and N represent the height and width of the image, M*N represents the total number of pixels in the image, ImgMean(x, y) represents the pixel value at the coordinate (x, y), and the summation symbol represents the summation of each pixel in the image. All pixel values ​​will be accumulated. Calculating the average value of multiple low-frequency components can effectively suppress random noise and offset the random noise in each frame, thereby improving the signal-to-noise ratio of the image;

[0057] S4, using the corrected background image to correct each frame of the panoramic photoelectric radar;

[0058] When correcting each frame of the panoramic photoelectric radar using the corrected background image, the histogram equalization and contrast enhancement are combined to enhance the image features. The correction process uses fast Fourier transform to accelerate the image processing speed. The correction formula is as follows:

[0059] ImgDst=ImgSrc-ImgBase

[0060] The corrected background image provides a stable benchmark, so that each frame of the image can be adjusted relative to the same standard, thereby reducing the image brightness fluctuation caused by changes in ambient light and ensuring the consistency of the image in the same scene. ImgDst represents the image obtained after correction processing, ImgSrc represents the original frame image, which refers to the current frame image obtained from the panoramic photoelectric radar, and ImgBase represents the corrected background image. The corrected background image can effectively eliminate artifacts caused by camera characteristics and other unavoidable factors, so that the obtained image quality is higher and more suitable for subsequent analysis and decision-making;

[0061] S5, stitching and displaying the corrected images;

[0062] The formula for stitching the rectified images is as follows:

[0063]

[0064] The corrected images usually have higher contrast and clarity. Stitching these images can better preserve details and achieve higher image quality, which is convenient for subsequent analysis and processing. In the formula, WarImg(x', y') represents the corresponding points in the stitched image, (x, y) represents the points in the original image, and H represents the homography matrix. Affine transformation is performed on each ImgDst. Since the stitching is performed on the corrected image, the correction result helps reduce the brightness inconsistency during stitching and the impact of other visual seams, making the final synthesized image more natural and seamless. By stitching multiple frames of corrected images, the acquired data can be fully utilized to improve the utilization rate of information.

[0065] The formula for stitching and fusing the rectified images is as follows:

[0066] ImgResult(x, y)=α*Img1(x, y)+(1-α)*Img2(x, y)

[0067] Through image fusion, the information of multiple frames can be integrated to further enhance the detection and recognition capabilities of the target, making the target easier to identify in a complex background. In the formula, ImgResult(x, y) represents the fused image at position (x, y), that is, the final output image, Img1(x, y) represents the pixel value of the first input image, Img2(x, y) represents the pixel value of the second input image, α represents the weight coefficient, and the value range is [0-1], which is used to control the proportion of the first image Img1 in the fusion. By fusing the corrected images, time series analysis can be effectively performed to facilitate the detection of changes and anomalies therein, providing support for dynamic monitoring and early warning.

[0068] By performing image subtraction processing on the real-time video and utilizing the non-repetitive imaging characteristics of the field of view images, the background image with uneven illumination is calculated to make the illumination uniform, thereby improving the continuity of the overall image effect.

[0069] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for processing brightness consistency of panoramic photoelectric radar imaging, characterized in that: The following steps are involved: S1. When the system is running, a medium-wave cooled infrared camera is used to collect each frame of the image, and a low-pass filtering process is performed before stitching. A low-pass filtering algorithm is used to obtain a low-frequency image Img(i); S2, summing up 50 consecutive frames of low-frequency images of the panoramic view and then calculating the average value; S3, normalizing the mean image, returning the average value of the low-frequency component to 0, and obtaining a corrected background image; S4, using the corrected background image to correct each frame of the panoramic photoelectric radar; S5. Stitching and displaying the corrected images.

2. The method for processing brightness consistency of panoramic photoelectric radar imaging according to claim 1, characterized in that: The S1 is low-pass filtered before stitching. The formula for obtaining the low-frequency image Img(i) using a low-pass filtering algorithm is as follows: ImgLF(i)=Filter2(Img(i)) In the formula, ImgLF(i) represents the low-frequency image of the i-th frame after low-pass filtering, Img(i) represents the image collected by the i-th frame, and Filter2 represents the operation of applying a two-dimensional filter to the input image.

3. The method for processing brightness consistency of panoramic photoelectric radar imaging according to claim 2, characterized in that: The formula for summing the continuous 50 frames of low-frequency images of the panoramic view in S2 is as follows: In the formula, ImgSum represents the sum of 50 frames of low-frequency images, ImgLF(i) represents the low-frequency image of the i-th frame after low-pass filtering, and i represents the number of image frames. The summation symbol indicates that the low-frequency images from the 1st frame to the 50th frame are summed.

4. The method for processing brightness consistency of panoramic photoelectric radar imaging according to claim 3, characterized in that: The formula for calculating the average value of 50 consecutive frames of low-frequency images of the panoramic view by S2 is as follows: ImgMean=ImgSum / 50 In the formula, ImgMean represents the average value of 50 consecutive frames of low-frequency images of the periscopic view, and ImgSum represents the sum of 50 frames of low-frequency images.

5. The method for processing brightness consistency of panoramic photoelectric radar imaging according to claim 4, characterized in that: The formula for calculating the average value of the low-frequency component in S3 is as follows: In the formula, It represents the average value of the low-frequency component, M and N represent the height and width of the image, M*N represents the total number of pixels in the image, ImgMean(x, y) represents the pixel value at the coordinate (x, y), and the summation symbol represents the summation of each pixel in the image, and all pixel values ​​will be accumulated.

6. The method for processing brightness consistency of panoramic photoelectric radar imaging according to claim 5, characterized in that: The S3 performs normalization processing on the mean image, and the average value of the low-frequency component is returned to 0. The formula for obtaining the corrected background image is as follows: In the formula, ImgBase represents the corrected background image, ImgMean represents the average value of 50 consecutive low-frequency images of the surrounding view, Represents the average value of low-frequency components.

7. The method for processing brightness consistency of panoramic photoelectric radar imaging according to claim 1, characterized in that: When the S4 uses the corrected background image to correct each frame of the panoramic photoelectric radar, it combines histogram equalization and contrast enhancement to enhance the image features. The correction process uses fast Fourier transform to accelerate the image processing speed.

8. The method for processing brightness consistency of panoramic photoelectric radar imaging according to claim 7, characterized in that: The formula for correcting each frame of the panoramic photoelectric radar by using the corrected background image in S4 is as follows: ImgDst=ImgSrc-ImgBase In the formula, ImgDst represents the image obtained after correction processing, ImgSrc represents the original frame image, which refers to the current frame image obtained from the panoramic photoelectric radar, and ImgBase represents the corrected background map.

9. The method for processing brightness consistency of panoramic photoelectric radar imaging according to claim 8, characterized in that: The formula for stitching the corrected images in S5 is as follows: In the formula, WarImg(x',y') represents the corresponding points in the spliced ​​image, (x, y) represents the points in the original image, H represents the homography matrix, and affine transformation is performed on each ImgDst.

10. A method for processing brightness consistency of panoramic photoelectric radar imaging according to claim 9, characterized in that: The formula for stitching and fusing the corrected images in S5 is as follows: ImgResult(x,y)=α*Img1(x,y)+(1-α)*Img2(x,y) In the formula, ImgResult(x,y) represents the fused image at position (x,y), that is, the final output image, Img1(x,y) represents the pixel value of the first input image, Img2(x,y) represents the pixel value of the second input image, and α represents the weight coefficient, which ranges from [0-1] and is used to control the proportion of the first image Img1 in the fusion.

Images