Image compression method and device suitable for star sensor

Abstract

The invention discloses an image compression method and device suitable for a star sensor, and belongs to the field of photoelectric imaging.The method comprises the steps that a low-frequency image is generated according to star sensor image compression, and a high-frequency image is generated based on high-pass filtering and Robinson Guard filtering according to the star sensor image; according to the high-frequency image, stroke coding is carried out on gray zero pixels, Huffman variable-length coding is carried out on non-zero pixels, and high-frequency image data are obtained; and generating compressed image data according to the low-frequency image data, the high-frequency image data and the Huffman table of the high-frequency compression in combination with the size information of the original image. According to the method, all useful information of the star sensor image is completely reserved while the calculation amount is reduced, image background noise is removed through filtering before high-frequency image coding, the coding accuracy is improved, stroke coding and Huffman variable-length coding are adopted for different pixels of high-frequency image data respectively, the data space is further saved, and the coding efficiency is improved. Therefore, the effects of small calculation amount, large compression ratio and high image restoration degree are achieved.

Description

Technical Field

[0001] The present invention relates to the field of photoelectric imaging and high-precision measurement technology, and in particular to an image compression method and device suitable for a star sensor. Background Art

[0002] A star sensor is a fixed-focus lens camera that can be used to take star maps. It identifies stars by extracting star points to obtain the position vector of the star image in the star sensor coordinate system, and can calculate the current attitude of the star sensor based on the position vector of the star point in the star sensor coordinate system and the position vector in the inertial system. Star sensors are mainly used on spacecraft and are difficult to recover. Obtaining the original image of the star sensor through telemetry and fully retaining the useful information in the image is an important means of post-analysis of various test experiments.

[0003] Star sensor images are usually megapixel (1 million to 4 million) grayscale images, with single pixel grayscale represented by 10 to 14 bits, and single image data volume of 2MByte to 8MByte. Due to communication channel bandwidth issues, the telemetry download speed of images in the experiment is generally only a dozen KBytes. At this time, it is very important to maximize the compression of images while fully retaining useful information.

[0004] Commonly used image compression algorithms include JPEG and JPEG2000 compression coding, which compress and encode the DCT transform and wavelet transform coefficients of image blocks respectively. DCT transform is prone to block effects, wavelet transform has a large amount of calculation, and under the condition of large compression ratio (>25), the distortion of compressed images of both transforms is serious. They are not suitable for telemetry transmission of star sensor images with high compression rate requirements, small calculation requirements, and high real-time requirements. Summary of the invention

[0005] The main purpose of the present invention is to solve the problems of image compression transmission and compressed image distortion under bandwidth-limited conditions such as star sensor images.

[0006] The invention provides an image compression method suitable for a star sensor, comprising: generating a low-frequency image according to star sensor image compression, and generating a high-frequency image according to the star sensor image based on high-pass filtering and Robinson-Guard filtering processing; performing run-length coding on grayscale zero pixels and Huffman variable-length coding on non-zero pixels according to the high-frequency image to obtain high-frequency image data; and generating compressed image data according to the low-frequency image data, the high-frequency image data, and a Huffman table of high-frequency compression in combination with original image size information.

[0007] In the image compression method applicable to the star sensor of the present invention, the generating of the low-frequency image according to the star sensor image compression includes: dividing the star sensor image into a plurality of closely adjacent sub-images, taking the average of each sub-image to generate pixels of a reduced low-frequency image, and obtaining the low-frequency image.

[0008] In the image compression method applicable to a star sensor of the present invention, the high-frequency image is generated based on the star sensor image and high-pass filtering and Robinson_Guard filtering, comprising: convolving the star sensor image with a high-pass filtering template to obtain a first intermediate image; performing a target enhancement operation on the first intermediate image to obtain a second intermediate image; and performing Robinson_Guard filtering on the second intermediate image to obtain the high-frequency image.

[0009] In the image compression method applicable to the star sensor of the present invention, in the process of convolving the star sensor image with a high-pass filter template, the high-pass filter template is a 2N+1 dimensional matrix, the outermost circle of the matrix is ​​equal negative values, the sum of the elements in the N+1th row and N+1th column and the outermost circle values ​​is 0, and the remaining elements are all 0, and the matrix dimension is determined according to the maximum value of the star target spot occupying the pixel.

[0010] In the image compression method applicable to the star sensor of the present invention, the high-pass filter template comprises:

[0011]

[0012] In the image compression method applicable to the star sensor of the present invention, the target enhancement operation on the first intermediate image includes: using an enhanced convolution template to perform a convolution operation on the first intermediate image, the enhanced convolution template is an N-dimensional matrix, the dimension is determined according to the minimum value of the star target spot occupying the pixel, and each element in the enhanced convolution template is equal.

[0013] In the image compression method applicable to the star sensor of the present invention, the enhanced convolution template comprises:

[0014]

[0015] In the image compression method applicable to the star sensor of the present invention, the grayscale zero pixel is run-length coded and the non-zero pixel is Huffman variable-length coded according to the high-frequency image, which includes: establishing a Huffman tree and a Huffman variable-length code table according to the probability of occurrence of pixel grayscale in the image; sequentially scanning the image pixels, for the grayscale non-zero pixel, variable-length coding is performed according to the one-to-one correspondence between the pixel grayscale and the Huffman variable-length code, and run-length coding is used for the grayscale zero pixel.

[0016] The present invention also provides an image compression device suitable for a star sensor, comprising: a low-frequency compression module, used for compressing a star sensor image to generate a low-frequency image; a high-frequency processing module, used for generating a high-frequency image based on the star sensor image and high-pass filtering and Robinson-Guard filtering; the high-frequency compression module, used for performing run-length coding on grayscale zero pixels and Huffman variable-length coding on non-zero pixels according to the high-frequency image to obtain high-frequency image data; and a comprehensive processing module, used for generating compressed image data according to the low-frequency image data, the high-frequency image data, the Huffman table of high-frequency compression, and the size information of the original image.

[0017] The present invention also provides a computer storage medium, which stores a computer program executable by a processor, and the computer program executes any one of the above-mentioned image compression methods applicable to a star sensor.

[0018] The beneficial effects of the present invention are as follows: by compressing the star sensor image into low-frequency images and high-frequency images respectively, the amount of calculation is reduced while all the useful information of the star sensor image is completely retained; before encoding the high-frequency image, the noise in the image background is removed by Robinson-Guard filtering to improve the encoding accuracy; and the run length coding and Huffman variable length coding are respectively used for different pixels of the high-frequency image data to further save data space, thereby achieving the effects of small amount of calculation, large compression ratio and high image restoration, and being very suitable for telemetry transmission of star sensor images. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 is a flow chart of an image compression method applicable to a star sensor according to an embodiment of the present invention;

[0021] Figure 2 is a grayscale histogram of a high-frequency image according to an embodiment of the present invention;

[0022] Figure 3 is one of the schematic diagrams of grayscale gradient and intensity of a low-frequency image according to an embodiment of the present invention;

[0023] Figure 4 This is the second schematic diagram of the grayscale gradient and intensity of a low-frequency image according to an embodiment of the present invention;

[0024] Figure 5 is a schematic diagram of image segmentation according to an embodiment of the present invention;

[0025] Figure 6 is a schematic diagram of the high-pass filtering processing result of a star image according to an embodiment of the present invention;

[0026] Figure 7 is a schematic diagram of star image enhancement filtering according to an embodiment of the present invention;

[0027] Figure 8 It is a Robinson Guard filtered image according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0029] The telemetry of the original image information of the star sensor requires a small amount of calculation, does not occupy too many computing resources of the star sensor, and has a high image compression ratio. By analyzing the image information of the star sensor, the present invention designs a fast and simple star sensor image compression method, divides the star sensor image into a low-frequency image and a high-frequency image, and then compresses the high-frequency information and the low-frequency information of the image using different compression methods. The image compression algorithm completely retains all the useful information of the star sensor image, has a small amount of calculation and a large compression ratio, and is very suitable for telemetry image transmission during the experiment.

[0030] According to the characteristics of the photoelectric image, a photoelectric image containing a star target can be described as:

[0031] f(x,y)=f T (x,y)+f N (x,y)+f B (x,y)

[0032] In the formula, (x, y) is the coordinate of the pixel point in the image; f(x, y) represents the gray value of the image; f T (x, y) represents the gray value of the target point; f N (x, y) represents the gray value of the noise point; f B (x,y) represents the grayscale value of the background image.

[0033] Background image B (x, y) is a large area with a slowly changing scene. There is a strong correlation between the grayscale of pixels, which occupies the low-frequency component of the image spatial frequency. T The brightness of (x,y) is higher than the background and is not related to the background. It is an isolated bright spot in the image and is the high-frequency part of the image. N(x, y) is the sum of all kinds of noise generated inside the sensor. It is unrelated to the background image and is the high-frequency part of the image. It can be approximated to have a mean of 0 and a variance of σ. 2 Gaussian white noise, its distribution in space is random, f N The spatial distribution of (x,y) has no correlation.

[0034] Combine the following Figures 1 to 8 The image compression method and device applicable to the star sensor of the present invention are described. Figure 1 FIG. 1 is a flow chart of an image compression method applicable to a star sensor according to an embodiment of the present invention. Figure 1 As shown, the present invention provides an image compression method suitable for a star sensor, comprising:

[0035] S1. Generate a low-frequency image according to star sensor image compression, and generate a high-frequency image according to the star sensor image based on high-pass filtering and Robinson_Guard filtering.

[0036] The star sensor mainly focuses on the precise position and brightness of the captured stars (small bright spots of stars) in the image. This target is the high-frequency part of the image. The low-frequency background of the star sensor represents the non-parallel stray light information during shooting, which is not the main focus. It is only necessary to understand the overall gradient and intensity information of the background to confirm the direction and intensity of the stray light during shooting. The high-frequency background noise of the star sensor is useless information and does not need to be paid attention to.

[0037] According to the focus of the star sensor image, the present invention first divides the star sensor image into a low-frequency image and a high-frequency image, and then compresses the high-frequency information and the low-frequency information of the image using different compression methods. Among them, the low-frequency image can be directly generated by image compression.

[0038] S2. According to the high-frequency image, run-length coding is performed on the grayscale zero pixels, and Huffman variable-length coding is performed on the non-zero pixels to obtain high-frequency image data.

[0039] High-frequency image information includes star information and image background noise. Image background noise is useless information and is not needed, while star information needs to be preserved as losslessly as possible. Huffman coding is the basis of data compression. Huffman coding is a lossless coding method that can fully obtain information in the image. Huffman coding only counts the distribution of image pixel grayscale to quickly obtain the Huffman variable-length coding table, with low computational complexity. High-frequency images are suitable for encoding using Huffman coding.

[0040] Although Huffman coding is a variable-length code, the compression ratio is still low for general images (where the pixel grayscale distribution is uniform). The compression ratio will be greatly improved only when encoding special images where the pixel grayscale distribution is relatively concentrated. The star sensor photographs stars, which are extremely weak small targets. The grayscale of the star points is not high, while the grayscale of the high-frequency image background is close to 0. Therefore, the grayscale distribution of the high-frequency image is relatively concentrated. The grayscale histogram of the high-frequency image can be seen. Figure 2 , the image grayscale is concentrated below 50, which is suitable for using Huffman coding to achieve a high compression ratio.

[0041] The high-frequency image of the star sensor is quite special. Many grayscale 0 pixels are connected. The code length of grayscale 0 is 1 bit. At this time, the grayscale 0 pixel adopts the run encoding method: 0 code word + 0 length (scanning along the image, the number of consecutive 0s) (a total of 6 bits). Each code represents up to 31 consecutive 0s, which can save a lot of data space. Grayscale non-0 pixels are directly encoded with Huffman variable-length code.

[0042] S3. Generate compressed image data according to the low-frequency image data, the high-frequency image data, the high-frequency compressed Huffman table, and the original image size information.

[0043] The low-frequency and high-frequency compressed image data are obtained, and the two are combined to obtain the entire compressed image data. The image width, height information and Huffman table are added in front of the data. The specific format is shown in Table 1.

[0044] Table 1

[0045] Serial number name Data length 1 Image width 2 bytes 2 Figure High 2 bytes 3 Huffman table 1280 bytes 4 Low frequency compressed image data …… 5 High frequency compressed image data ……

[0046] Using this compression method, the image compression ratio can reach an average of 35 to 40. At the same time, the information that the star sensor is concerned about is completely retained without occupying too many star sensor computing resources. It is very suitable for star sensor image compression telemetry downlink for data analysis.

[0047] The image compression method applicable to star sensors of the present invention reduces the amount of calculation while completely retaining all useful information of the star sensor images by compressing the star sensor images according to low-frequency images and high-frequency images respectively. Noise in the image background is removed by Robinson-Guard filtering before encoding to improve encoding accuracy. In combination with using run length coding and Huffman variable length coding for different pixels of high-frequency image data respectively, data space is further saved, thereby achieving the effects of small amount of calculation, large compression ratio and high image restoration degree. The method is very suitable for telemetry transmission of star sensor images.

[0048] In one embodiment, generating a low-frequency image based on star sensor image compression includes: dividing the star sensor image into a plurality of closely adjacent sub-images, taking an average of each sub-image to generate pixels of a reduced low-frequency image, and obtaining the low-frequency image.

[0049] The low-frequency image compression of the star sensor directly reduces the image by interpolation, and then uses image interpolation to enlarge it to the original image size during decoding. That is, for the low-frequency part, it is only necessary to output the slowly changing grayscale gradient in the image and the grayscale mean (intensity) of each block.

[0050] For example, the original image is divided into closely adjacent 16×16 sub-image blocks, each of which is generated as a pixel of a reduced low-frequency image, and the pixel grayscale is the average value of the 16×16 block pixels. The calculation is shown in the formula

[0051] Formula 1. The size of the low-frequency compressed data becomes the original low-frequency image

[0052]

[0053] like Figure 3 As shown, the grayscale gradient points from the lower right corner to the upper left corner, and the grayscale intensity is strongest in the upper left corner and weakest in the lower right corner. Figure 3 The actual size is 768×572. Figure 4 The image is interpolated and reduced to 48×36, and the grayscale gradient and intensity information can also be fully understood. For specific image blocks, see Figure 5 Low-frequency image compression directly reduces the image by interpolation, and then uses image interpolation to enlarge it to the original image size during decoding.

[0054] Among them, the low-frequency image should be the low-frequency part of the image extracted by convolution of a low-pass filter. Since the average of the 16×16 blocks is a low-pass filter, the compressed image is a low-frequency image. Therefore, the original image is directly reduced here, saving the calculation amount of low-pass filtering.

[0055] In one embodiment, the generating a high-frequency image based on the star sensor image and high-pass filtering and Robinson_Guard filtering includes: convolving the star sensor image with a high-pass filtering template to obtain a first intermediate image; performing a target enhancement operation on the first intermediate image to obtain a second intermediate image; and performing Robinson_Guard filtering on the second intermediate image to obtain the high-frequency image.

[0056] According to the characteristics of star sensors, in order to improve the coding compression rate, a special high-pass filtering algorithm is used when extracting high-frequency images from star sensor images to highlight the star target information in the high-frequency images, while suppressing the high-frequency information of the noise part as much as possible to ensure that a large number of pixels in the image have a grayscale of 0. The high-frequency image generation algorithm for star sensors is divided into three steps: 1) High-pass filter convolution removes stray light background in the image. Figure 6 This is the effect of a strong stray light image after high-pass filtering. After image processing, the stray light is completely removed, and only the star target and background high-frequency noise remain in the star map. 2) Simple template filtering target enhancement. 3) Robinson_Guard filtering suppresses background noise. These three steps are the preprocessing process for extracting star targets in the image.

[0057] In one embodiment, in the process of convolving the star sensor image with a high-pass filter template, the high-pass filter template is a 2N+1 dimensional matrix, the outermost circle of the matrix is ​​equal negative values, the sum of the elements in the N+1th row and N+1th column and the outermost circle is 0, and the remaining elements are all 0, and the matrix dimension is determined according to the maximum value of the star target spot occupying the pixel.

[0058] On the one hand, the high-pass filter needs to remove the low-frequency background, and on the other hand, it needs to completely retain the energy of the star target. The commonly used small template high-pass filter will also enhance the high-frequency noise when removing the background. According to the low-frequency characteristics of the background signal of the star map, the embodiment of the present invention designs a 2N+1)×(2N+1) edge high-pass filter template to convolve the image, which can filter out the low frequency well and completely retain the star target in the image without enhancing the high-frequency noise. Among them, 2N+1 is consistent with the maximum value of the star target spot occupying the pixel.

[0059] In one embodiment, the high-pass filter template includes:

[0060]

[0061] Considering that the maximum value of the star target spot in pixels is usually 13×13, a 13×13 edge high-pass filter template can be designed to convolve the image, see Formula 2. This template can filter out low frequencies well and completely retain the star targets in the image without enhancing high-frequency noise.

[0062] In one embodiment, the target enhancement operation on the first intermediate image includes: using an enhanced convolution template to perform a convolution operation on the first intermediate image, the enhanced convolution template is an N-dimensional matrix, the dimension is determined according to the minimum value of the star target spot occupying the pixel, and each element in the enhanced convolution template is equal.

[0063] Under strong stray light conditions, the variance of the background high-frequency noise σ is very large, and the grayscale of each pixel in the star target spot may not meet the requirement of >3σ variance during normal detection. Target enhancement is to increase the grayscale difference between the star target and the background pixels by grayscale stretching.

[0064] However, if the pixel grayscale of the star target area and the background area is stretched synchronously by the magnification, the signal-to-noise ratio of the star target will not change, which is not helpful for the detection of weak targets. Taking advantage of the fact that the local energy of the star target imaging spot is concentrated while the background high-frequency noise is random, the present invention designs a simple target enhancement filter and strengthens the target by energy aggregation.

[0065] The image is convolved with an enhanced convolution template, which is similar to a low-pass smoothing template, but is not divided by a coefficient to avoid truncation loss of the target area energy due to division. After filtering, the grayscale peak value of the star target area is equivalent to the energy sum of all pixels of the star target. The background noise is random, and although it also increases the background variance after filtering, the increase is lower than that of the target, and the signal-to-noise ratio of the star target is increased.

[0066] In one embodiment, the enhanced convolution template includes:

[0067]

[0068] Assuming that the star target spot occupies 3×3 pixels, the signal-to-noise ratio is increased by 1 times by using the above template filtering. Figure 6 It is an image with target enhancement, where very weak targets appear clearly.

[0069] In one embodiment, the method of performing run length coding on zero grayscale pixels and Huffman variable length coding on non-zero pixels based on the high-frequency image includes: establishing a Huffman tree and a Huffman variable length code table based on the probability of occurrence of pixel grayscale in the image; sequentially scanning the image pixels, performing variable length coding on non-zero grayscale pixels based on the one-to-one correspondence between the pixel grayscale and the Huffman variable length code, and using run length coding on zero grayscale pixels.

[0070] Methods for suppressing stray light background include: using nonlinear nonparametric Robinson Guard filter algorithm to suppress background noise. Robinson Guard filter is a background suppression filter based on guard band. The filter template is shown in the following formula. The filter output value of the center point (i, j) is calculated through the pixels outside the guard band.

[0071]

[0072] Among them, f pre (i,j) represents the original pixel grayscale at (i,j), f Robinson(i, j) represents the Robinson Guard filter output. Here, the Robinson Guard filter template is selected as 13×13 according to the size of the star target and the angular velocity of the motion.

[0073] The star sensor high-frequency image compression can adopt the conventional Huffman coding method, and the Huffman tree and Huffman variable-length code table are established according to the probability of pixel grayscale in the image. The grayscale of the star sensor image is represented by 10 to 12 bits, but the grayscale of high-frequency image pixels is actually concentrated below 50. The Huffman table can only count the grayscale information of 0 to 255, and a very small number of pixels with grayscale > 255 can be treated as 255, which does not affect the acquisition of useful information.

[0074] Scan the image pixels sequentially from the upper left corner to the lower right corner, and implement variable-length code compression based on the one-to-one correspondence between pixel grayscale and Huffman code. Grayscale 0 pixels use run-length encoding.

[0075] The present invention also provides an image compression device for a star sensor, comprising: a low-frequency compression module, used for compressing a star sensor image to generate a low-frequency image; a high-frequency processing module, used for generating a high-frequency image based on the star sensor image and high-pass filtering and Robinson-Guard filtering; the high-frequency compression module, used for performing run-length coding on grayscale zero pixels and Huffman variable-length coding on non-zero pixels according to the high-frequency image to obtain high-frequency image data; and a comprehensive processing module, used for generating compressed image data according to the low-frequency image data, the high-frequency image data, the Huffman table of high-frequency compression, and the size information of the original image.

[0076] The image compression device for a star sensor provided in the embodiment of the present invention has the same implementation principle and technical effects as those of the aforementioned star sensor image compression method embodiment. For the sake of brief description, for matters not mentioned in the embodiment of the image compression device for a star sensor, reference may be made to the corresponding contents in the aforementioned star sensor image compression method embodiment.

[0077] The present application also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a disk, an optical disk, a server, an App application store, etc., on which a computer program is stored, and when the program is executed by a processor, a corresponding function is realized. When the computer-readable storage medium of this embodiment is executed by a processor, the method embodiment is applicable to an image compression method for a star sensor.

[0078] It should be pointed out that, according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, and two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0079] The order of execution of each step in the above embodiment does not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

[0080] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims

1. An image compression method suitable for a star sensor, characterized in that: include: Generate a low-frequency image according to the star sensor image compression, and generate a high-frequency image according to the star sensor image based on high-pass filtering and Robinson-Guard filtering processing; According to the high-frequency image, the grayscale zero pixels are run-length coded, and the non-zero pixels are Huffman variable-length coded to obtain the high-frequency image data; Compressed image data is generated based on low-frequency image data, high-frequency image data, and the high-frequency compressed Huffman table, combined with the original image size information.

2. The image compression method suitable for star sensor according to claim 1, characterized in that: The step of generating a low-frequency image according to the star sensor image compression comprises: The star sensor image is divided into a plurality of closely adjacent sub-images, and each sub-image is averaged to generate pixels of a reduced low-frequency image, thereby obtaining a low-frequency image.

3. The image compression method suitable for star sensor according to claim 1, characterized in that: The step of generating a high-frequency image based on the star sensor image and high-pass filtering and Robinson_Guard filtering comprises: Convolving the star sensor image with a high-pass filter template to obtain a first intermediate image; performing a target enhancement operation on the first intermediate image to obtain a second intermediate image; The second intermediate image is processed by Robinson_Guard filtering to obtain the high-frequency image.

4. The image compression method suitable for star sensor according to claim 3, characterized in that: In the process of convolving the star sensor image with a high-pass filter template, the high-pass filter template is a 2N+1 dimensional matrix, the outermost circle of the matrix is ​​equal negative values, the sum of the elements in the N+1th row and N+1th column and the outermost circle is 0, and the remaining elements are all 0, and the matrix dimension is determined according to the maximum value of the star target spot occupying the pixel.

5. The image compression method applicable to a star sensor according to claim 3 or 4, wherein the high-pass filter template comprises:

6. The image compression method suitable for star sensor according to claim 3, characterized in that: The performing a target enhancement operation on the first intermediate image includes: The first intermediate image is convolved using an enhanced convolution template, where the enhanced convolution template is an N-dimensional matrix, the dimension of which is determined according to the minimum value of the star target spot occupying the pixels, and each element in the enhanced convolution template is equal.

7. The image compression method suitable for star sensor according to claim 6, characterized in that: The enhanced convolution template includes:

8. The image compression method suitable for star sensor according to claim 1, characterized in that: The method of performing run length coding on grayscale zero pixels and Huffman variable length coding on non-zero pixels according to the high frequency image includes: Establish a Huffman tree and a Huffman variable-length code table according to the probability of occurrence of pixel grayscale in the image; The image pixels are scanned sequentially. For non-zero grayscale pixels, variable-length coding is performed based on the one-to-one correspondence between pixel grayscale and Huffman variable-length code. For zero grayscale pixels, run-length coding is used.

9. An image compression device suitable for a star sensor, characterized in that: include: A low-frequency compression module, used to compress the star sensor image to generate a low-frequency image; A high-frequency processing module, used for generating a high-frequency image based on the star sensor image by high-pass filtering and Robinson-Guard filtering; A high-frequency compression module is used to perform run-length coding on grayscale zero pixels and Huffman variable-length coding on non-zero pixels according to the high-frequency image to obtain high-frequency image data; The comprehensive processing module generates compressed image data according to the low-frequency image data, the high-frequency image data, the high-frequency compressed Huffman table and the original image size information.

10. A computer storage medium, characterized in that: A computer program executable by a processor is stored therein, and the computer program executes the image compression method applicable to a star sensor as claimed in any one of claims 1 to 8.

Images