A compensation method for low-grayscale and narrow-dynamic-range imaging data from optical satellite payloads

Abstract

The present invention addresses the characteristics of low overall grayscale values ​​(DN) and small dynamic range and entropy values ​​in the original payload images of remote sensing optical satellites during on-orbit payload imaging missions, caused by poor solar illumination conditions at the time of imaging. The present invention provides a method for compensating for low grayscale and narrow dynamic range imaging data from optical satellite payloads, including: spatial preprocessing of the original payload image data before onboard compression; onboard compression processing of the preprocessed image data to obtain compressed code stream data and downlink it; and decompression and post-processing of the compressed code stream data downlinked from the satellite on the ground to obtain the final decompressed and restored image. The present invention adds an adaptive level shifting preprocessing algorithm before onboard image compression processing to compensate for the original payload imaging data and adaptively increase the image grayscale value, making the payload image characteristics more favorable for wavelet quantization coding in the compression algorithm, further improving compression performance.

Description

Technical Field

[0001] This invention relates to the field of optical remote sensing satellite data processing and transmission systems, and is applicable to the design of satellite systems that are equipped with optical payloads and realize on-board compression functions. Background Technology

[0002] The different local times of the descending node and the varying solar altitude angles of optical remote sensing satellites result in different levels of sunlight exposure and reflection from ground objects. Consequently, the light intensity captured by the onboard optical detectors also affects the characteristics of the imaging data. Raw onboard image data acquired by payloads with poor illumination conditions at the time of imaging exhibits concentrated DN values, a narrow dynamic range, and low entropy, which is detrimental to wavelet quantization encoding in onboard compression algorithms.

[0003] Taking 10-bit as an example, when the dynamic range of the DN value of more than 80% of the pixels in the image is within 300, it will cause the wavelet quantization coding performance of the satellite compression algorithm to degrade, resulting in poor image quality after ground decompression and recovery. Summary of the Invention

[0004] The technical problem solved by this invention is to overcome the shortcomings of the prior art and provide an on-orbit dynamic compensation method that greatly reduces the amplitude of wavelet coefficients, reduces the number of planes that need to be encoded, and the preprocessed data distribution is more conducive to entropy coding, thereby further improving compression performance.

[0005] The technical solution of this invention is: a compensation method for low grayscale and narrow dynamic range imaging data of optical satellite payloads, comprising:

[0006] Before on-board compression, the original image data of the payload is preprocessed in the spatial domain.

[0007] The preprocessed image data is compressed on-board to obtain compressed bitstream data, which is then transmitted down.

[0008] After receiving the compressed bitstream data transmitted from the satellite, the ground performs decompression and post-processing to obtain the final decompressed and restored image.

[0009] The spatial preprocessing includes: shifting the overall distribution of the DN values ​​of the original payload image to the center of the corresponding quantization bit range of the DN values ​​through adaptive level shifting, thereby changing the distribution characteristics of the DN values ​​of the original payload image and obtaining the preprocessed image.

[0010] The step of shifting the overall distribution of the original DN values ​​of the payload image to the center of the corresponding quantization bit range by adaptive level shifting includes: caching the original payload image data according to the image quantization bit parameters obtained through external instructions and the pre-set image compression block size; summing the cached original payload image data according to the pixel DN values ​​and dividing by the number of pixels to calculate the average DN value of the original payload image block; calculating the level shift value according to the image quantization bit and the average DN value of the original payload image block; and shifting the DN values ​​of the original payload image block by adding the level shift value to obtain the preprocessed image block.

[0011] Adaptive level shifting refers to adaptively determining the level shift value for each image based on the payload quantization bit depth, then performing a corresponding level shift on the entire image, changing the characteristics of the original satellite image data, and finally compressing the shifted image using the original satellite compression method.

[0012] The step of performing on-board compression processing on the preprocessed image data to obtain compressed bitstream data includes: compressing the preprocessed image data to generate initial compressed bitstream data; and inserting level shift values ​​at the positions agreed upon in the frame header format of the obtained initial compressed bitstream to obtain the final compressed bitstream data.

[0013] After receiving the compressed bitstream data transmitted from the satellite, the ground performs decompression and post-processing, including: decompressing the compressed bitstream data to obtain a decompressed image; extracting the level shift value from the position agreed upon in the initial compressed bitstream frame header format; subtracting the level shift value from the DN value of the decompressed image to obtain the final decompressed and recovered image.

[0014] When the DN value of the decompressed image is subtracted from the level shift value, if the calculation result is negative, the DN value of the calculation result is clamped to 0 to obtain the final decompressed and restored image.

[0015] The term "low grayscale" refers to the image's DN value range being less than 255.

[0016] The narrow dynamic range refers to the range of difference in the image DN values ​​within the compressed block area being between 100 and 200.

[0017] The advantages of this invention compared to the prior art are:

[0018] 1. Reduce the impact of illumination conditions on the distribution characteristics of the original image data of the optical payload satellite due to the local time of the descent intersection point in orbit, and improve the wavelet quantization coding performance of the compression algorithm;

[0019] 2. The image preprocessing process is simple and flexible to operate, without adding extra satellite configuration resources. Image spatial preprocessing is achieved through the internal software design of the already configured onboard hardware resources.

[0020] 3. It does not require changes to the original compression algorithm and protocol format, and has strong adaptability. It only needs to be completed on the ground, resulting in low computational complexity;

[0021] 4. The adaptive level shifting preprocessing algorithm is flexible and can adapt to the original compression method, mean adaptive level shifting method, median adaptive level shifting method, block mean adaptive level shifting method, and fixed level shifting method. This enables flexible expansion of system functions. Attached Figure Description

[0022] Figure 1 This is a system block diagram of the present invention. Detailed Implementation

[0023] like Figure 1 As shown, the present invention provides a method for compensating low grayscale and narrow dynamic range imaging data of optical satellite payloads, comprising:

[0024] Before on-board compression, the original image data of the payload is preprocessed in the spatial domain.

[0025] The preprocessed image data is compressed on-board to obtain compressed bitstream data, which is then transmitted down.

[0026] After receiving the compressed bitstream data transmitted from the satellite, the ground performs decompression and post-processing to obtain the final decompressed and restored image.

[0027] The spatial preprocessing includes: shifting the overall distribution of the DN values ​​of the original payload image to the center of the corresponding quantization bit range of the DN values ​​through adaptive level shifting, thereby changing the distribution characteristics of the DN values ​​of the original payload image and obtaining the preprocessed image.

[0028] The step of shifting the overall distribution of the original DN values ​​of the payload image to the center of the corresponding quantization bit range by adaptive level shifting includes: caching the original payload image data according to the image quantization bit parameters obtained through external instructions and the pre-set image compression block size; summing the cached original payload image data according to the pixel DN values ​​and dividing by the number of pixels to calculate the average DN value of the original payload image block; calculating the level shift value according to the image quantization bit and the average DN value of the original payload image block; and shifting the DN values ​​of the original payload image block by adding the level shift value to obtain the preprocessed image block.

[0029] Adaptive level shifting refers to adaptively determining the level shift value for each image based on the payload quantization bit depth, then performing a corresponding level shift on the entire image, changing the characteristics of the original satellite image data, and finally compressing the shifted image using the original satellite compression method.

[0030] The step of performing on-board compression processing on the preprocessed image data to obtain compressed bitstream data includes: compressing the preprocessed image data to generate initial compressed bitstream data; and inserting level shift values ​​at the positions agreed upon in the frame header format of the obtained initial compressed bitstream to obtain the final compressed bitstream data.

[0031] After receiving the compressed bitstream data transmitted from the satellite, the ground performs decompression and post-processing, including: decompressing the compressed bitstream data to obtain a decompressed image; extracting the level shift value from the position agreed upon in the initial compressed bitstream frame header format; subtracting the level shift value from the DN value of the decompressed image to obtain the final decompressed and recovered image.

[0032] When the DN value of the decompressed image is subtracted from the level shift value, if the calculation result is negative, the DN value of the calculation result is clamped to 0 to obtain the final decompressed and restored image.

[0033] The term "low grayscale" refers to the image's DN value range being less than 255.

[0034] The narrow dynamic range refers to the range of difference in the image DN values ​​within the compressed block area being between 100 and 200.

[0035] Taking the original image data of the payload with a 10-bit quantization width as an example, the specific implementation steps are as follows:

[0036] 1) Before on-board compression, calculate the mean m of each image. e ;

[0037] The level shift value s is:

[0038] s = 1024 - m, m = m e

[0039] 2) The original load image X is level-shifted according to the level shift value s to obtain the preprocessed image.

[0040] Like Y, it performs image spatial preprocessing, that is:

[0041] Y = X + s

[0042] 3) Compress the shifted image Y by a factor of n using 11 bits to obtain compressed bitstream data. Then, compress the s...

[0043] Write it to the location specified in the compressed bitstream format.

[0044] 4) At the ground decoding end, decompression is performed to obtain the decompressed image. Perform the corresponding reverse shift, i.e.

[0045]

[0046] Obtain the final decompressed and restored image

[0047] Compression tests were conducted on image sets of various types and characteristics (see Table 1 below) using the original compression method, fixed 1024 level shift, mean adaptive level shift, median adaptive level shift, and block mean adaptive level shift.

[0048] Table 1 Description of various test items

[0049]

[0050] Comprehensive test and analysis results: Adding image spatial preprocessing before compression has no impact on the quality of decompressed and restored images; the average PSNR (Peak Signal-to-Noise Ratio) is improved by 3dB, and the MSE (Mean Square Error) is reduced by 37.5%.

Claims

1. A method for compensating low grayscale and narrow dynamic range imaging data of an optical satellite payload, characterized in that, include: Before on-board compression, the original image data of the payload is preprocessed in the spatial domain. The preprocessed image data is compressed on-board to obtain compressed bitstream data, which is then transmitted down. After receiving the compressed bitstream data transmitted from satellite to ground, the ground performs decompression and post-processing to obtain the final decompressed and restored image. By adaptive level shifting, the overall distribution of the DN values ​​of the original payload image is shifted to the center of the corresponding quantization bit range. This includes: buffering the original payload image data based on the image quantization bit parameters obtained through external instructions and the pre-set image compression block size; summing the buffered original payload image data according to the pixel DN values ​​and dividing by the number of pixels to calculate the average DN value of the original payload image block; calculating the level shift value based on the image quantization bit and the average DN value of the original payload image block; and shifting the original payload image block by adding the level shift value to the DN value to obtain the preprocessed image block. Adaptive level shifting refers to adaptively determining the level shift value for each image based on the payload quantization bit depth, then performing a corresponding level shift on the entire image, changing the characteristics of the original satellite image data, and finally compressing the shifted image using the original satellite compression method.

2. The compensation method for low grayscale and narrow dynamic range imaging data of an optical satellite payload according to claim 1, characterized in that: The spatial preprocessing includes: shifting the overall distribution of the DN values ​​of the original payload image to the center of the corresponding quantization bit range of the DN values ​​through adaptive level shifting, thereby changing the distribution characteristics of the DN values ​​of the original payload image and obtaining the preprocessed image.

3. The compensation method for low grayscale and narrow dynamic range imaging data of an optical satellite payload according to claim 1, characterized in that: The step of performing on-board compression processing on the preprocessed image data to obtain compressed bitstream data includes: compressing the preprocessed image data to generate initial compressed bitstream data; and inserting level shift values ​​at the positions agreed upon in the frame header format of the obtained initial compressed bitstream to obtain the final compressed bitstream data.

4. The compensation method for low grayscale and narrow dynamic range imaging data of an optical satellite payload according to claim 1, characterized in that: After receiving the compressed bitstream data transmitted from the satellite, the ground performs decompression and post-processing, including: decompressing the compressed bitstream data to obtain a decompressed image; extracting the level shift value from the position agreed upon in the initial compressed bitstream frame header format; subtracting the level shift value from the DN value of the decompressed image to obtain the final decompressed and recovered image.

5. The compensation method for low grayscale and narrow dynamic range imaging data of an optical satellite payload according to claim 4, characterized in that: When the DN value of the decompressed image is subtracted from the level shift value, if the calculation result is negative, the DN value of the calculation result is clamped to 0 to obtain the final decompressed and restored image.

6. The compensation method for low grayscale and narrow dynamic range imaging data of an optical satellite payload according to claim 1, characterized in that: The term "low grayscale" refers to the image's DN value range being less than 255.

7. The compensation method for low grayscale and narrow dynamic range imaging data of an optical satellite payload according to claim 1, characterized in that: The narrow dynamic range refers to the range of difference in the image DN values ​​within the compressed block area being between 100 and 200.

Images