A sub-block segmentation method based on spaceborne remote sensing image compression

Abstract

The present application relates to a kind of sub-block segmentation methods based on spaceborne remote sensing image compression, belong to remote sensing image processing technical field.The method of the present application includes: the pre-processing operation is carried out to the received spaceborne remote sensing image data stream, judge whether the data after pre-processing is compressed mode, if not, directly output corresponding data according to the target required spectral range, if yes, the data after pre-processing is split into auxiliary data and image data two parts;For auxiliary data, after uniform processing according to fixed packet number, it is exported;For image data, according to the number of pixels, the size of compressed sub-block is determined to segment, and the segmented sub-block is sequentially input into compression chip, finally output compressed code stream block.The method of the present application trades off with the engineering advantage of no need to split spectral range, packet number uniform, low cache occupation, short pre-processing delay with very small image quality loss, is more suitable for the actual working condition of spaceborne platform resource limited, is an optimization scheme considering image quality reliability and engineering practicability.

Description

Technical Field

[0001] This invention relates to the field of remote sensing image processing technology, and specifically to a sub-tile segmentation method based on spaceborne remote sensing image compression. Background Technology

[0002] In recent years, the development of global Earth observation capabilities has continued to deepen. Remote sensing satellites, as the core technology carrier for acquiring space information, are not only an important indicator of a nation's aerospace technology development level, but also shoulder the significant strategic mission of safeguarding national security, resource exploration, and ecological monitoring. With the iterative upgrades of semiconductor manufacturing processes and high-performance optical devices, the resolution of my country's remote sensing payloads has broken through the sub-meter level, and the volume of single-scene image data is growing exponentially. In the current planetary-to-ground data transmission mechanism, satellites need to temporarily store the collected data in onboard storage units, and then transmit it down after entering the coverage area of ​​ground stations. Limited by the dual constraints of onboard storage unit capacity and satellite-to-ground communication bandwidth, the collaborative transmission of massive amounts of high-resolution remote sensing data faces the dual challenges of efficiency and reliability. Against this backdrop, developing real-time compression technology with low time consumption and low resource consumption characteristics has become a key technological path to overcome storage and transmission bottlenecks and expand remote sensing application scenarios.

[0003] Sub-tile segmentation, as a core pre-processing step in spaceborne remote sensing image compression, directly determines the memory usage, real-time performance, and reconstruction quality of the compression process, especially having a crucial impact on the collaborative processing of heterogeneous spectral bands in multispectral remote sensing scenes. Spaceborne remote sensing systems commonly employ TDI (Time Delay Integration) mode to improve the imaging signal-to-noise ratio. Constrained by this mode and data transmission requirements, full-spectral and multispectral band data are typically transmitted in a mixed transmission mode, output line by line. Due to the differences in imaging mechanisms between full-spectral and multispectral bands, their line frequencies inherently differ, leading to a mismatch in the number of data lines. The traditional sub-tile segmentation method involves receiving mixed spectral band line data, temporarily storing it in a buffer, restoring it to complete full-spectral / multispectral band data, performing sub-tile segmentation based on spectral bands, and generating heterogeneous sub-tiles. This traditional sub-tile segmentation method requires designing separate sub-tile segmentation strategies for different spectral bands, resulting in the simultaneous presence of heterogeneous sub-tiles across multiple spectral bands in the buffer, significantly increasing memory usage. In addition, traditional sub-plot segmentation methods require first restoring the spectral data of each segment from the row-by-row mixed output into complete single-segment data before performing block processing. This process not only prolongs the data preprocessing latency, but also requires additional cache space to store the restored complete spectral data, further exacerbating the resource burden on the spaceborne platform. Summary of the Invention

[0004] This invention aims to solve the technical problems of process redundancy and cache burden in traditional sub-tile segmentation methods in the prior art, and provides a sub-tile segmentation method based on spaceborne remote sensing image compression.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] A sub-tile segmentation method based on spaceborne remote sensing image compression includes the following steps:

[0007] Step 1: Perform preprocessing operations on the received spaceborne remote sensing image data stream, filtering and retaining the valid data portion;

[0008] Step 2: Determine whether the data after preprocessing in Step 1 is in compression mode. If not, output the corresponding data directly according to the target required spectral band. If so, split the preprocessed data into two parts: auxiliary data and image data.

[0009] Step 3: For auxiliary data, extract and assemble auxiliary data packets according to the data transmission order, process them uniformly with a fixed number of packets, and directly output the data;

[0010] Step 4: For image data, based on the data transmission order and the number of pixels, determine the size of the compressed sub-plot and perform row-separated sub-plot segmentation. If the remaining number of rows is less than the preset size, fill the remaining rows to ensure the uniformity of the sub-plot size.

[0011] Step 5: Input the sub-blocks from Step 4 into the compression chip in sequence, and finally output the compressed bitstream block.

[0012] In the above technical solution, step 1 specifically involves: filtering and retaining valid data portions of the received spaceborne remote sensing image data stream by judging the frame header, frame tail, and byte length.

[0013] In the above technical solution, the above-mentioned sub-tile segmentation method based on spaceborne remote sensing image compression is applicable to sub-tile segmentation of 9-band TDICCD image data.

[0014] In the above technical solution, in step 4, the image data is composed of a data packet consisting of 4 consecutive rows of full chromatographic segments and 8 spectral segments of multispectral segments, and is continuously transmitted along the data transmission path according to this data packet.

[0015] In the above technical solution, every 4 consecutive full chromatographic segments are labeled as P1~P4, and the 8-segment multispectral segments are labeled as B1~B8. Then each data packet is output in the following order: P1~B1~B2~P2~B3~B4~P3~B5~B6~P4~B7~B8.

[0016] In the above technical solution, in step 4: the full chromatographic segment and multispectral segment in each data packet are divided into sub-blocks according to the number of pixels of 1536, and a total of 24 sub-blocks are formed after the full chromatographic segment and multispectral segment data are divided.

[0017] In the above technical solution, step 3 specifically includes:

[0018] Using 128 data packets as a processing unit, the auxiliary data of all spectral bands are extracted sequentially according to the data transmission order of P1~B1~B2~P2~B3~B4~P3~B5~B6~P4~B7~B8 to form an auxiliary data packet, which is then directly output after extraction.

[0019] The beneficial effects of this invention are:

[0020] This invention presents a sub-tile segmentation method based on spaceborne remote sensing image compression. It employs a row-separated sub-tile segmentation strategy, which is well-suited to the row-by-row output characteristics of TDI mode. It eliminates restrictions on data transmission formats and can directly perform block processing on mixed-output row data, eliminating redundant steps of data rearrangement. This streamlines the preprocessing process and effectively reduces the data buffering burden on the compression unit. Furthermore, this strategy is compatible with multiple image formats, allowing flexible selection of spectral segments to be compressed based on task requirements, adapting to different remote sensing application scenarios. Through structured design, the number of rows in the full-spectral segment sub-tiles is perfectly matched with the number of multispectral segments, achieving collaborative compression processing of multiple spectral segments within the same unit and process. This fundamentally solves the core challenge of collaborative processing of heterogeneous spectral segments, providing a new technical solution for high-performance, cost-effective remote sensing data storage and transmission.

[0021] The sub-tile segmentation method based on spaceborne remote sensing image compression of the present invention achieves engineering advantages such as no need to split spectral segments, uniform packet number, low cache usage, and short preprocessing latency with minimal image quality loss. It is more suitable for the actual working conditions of spaceborne platforms with limited resources and is an optimized solution that takes into account both image quality reliability and engineering practicality. Attached Figure Description

[0022] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0023] Figure 1 This is a flowchart illustrating the sub-tile segmentation method based on spaceborne remote sensing image compression according to the present invention.

[0024] Figure 2 This is a diagram of the data transmission format.

[0025] Figure 3 This is a diagram to illustrate the data packet format.

[0026] Figure 4 This is a schematic diagram of sub-block cutting using the traditional sub-block segmentation method.

[0027] Figure 5 This is a schematic diagram of sub-block cutting in the sub-block segmentation method of the present invention.

[0028] Figure 6Comparison of decompressed and restored images using different sub-plot segmentation methods, where (a) is the original image, (b) is the image after compression and decompression using the traditional sub-plot segmentation method, and (c) is the image after compression and decompression using the row-separated sub-plot segmentation method of the present invention. Detailed Implementation

[0029] The present invention will now be described in detail with reference to the accompanying drawings.

[0030] like Figure 1 As shown ( Figure 1 (Only a summary of the steps of the method of the present invention is shown in the figure). The present invention provides a sub-tile segmentation method based on spaceborne remote sensing image compression, which includes the following steps:

[0031] Step 1: Perform preprocessing operations on the received spaceborne remote sensing image data stream (data input), filtering and retaining the valid data portion; specifically:

[0032] The received satellite remote sensing image data stream is filtered and retained based on the frame header, frame tail, and byte length. This step is to ensure that the data format and content are correct and to exclude abnormal data.

[0033] Step 2: Determine whether the data after preprocessing in Step 1 is in compression mode. If not, output the corresponding data directly according to the target required spectral band. If so, split the preprocessed data into two parts: auxiliary data and image data.

[0034] Both the compression mode and the pass-through mode (i.e., uncompressed mode) of the data contain two parts: auxiliary data and image data. The auxiliary data is not compressed and its number of bytes is fixed. Only the image data participates in compression, and the auxiliary data is split according to the number of bytes.

[0035] Step 3: For auxiliary data, extract and assemble auxiliary data packets according to the data transmission order, process them uniformly with a fixed number of packets, and directly output the data;

[0036] Step 4: For image data, based on the data transmission order and the number of pixels, determine the size of the compressed sub-plot and perform row-separated sub-plot segmentation. If the remaining number of rows is less than the preset size, fill the remaining rows to ensure the uniformity of the sub-plot size.

[0037] The preset size refers to the height of the compressed sub-tile. Assuming the preset size height is 128, if the remaining number of rows is less than 128, FF fill will be used to fill the remaining rows.

[0038] Step 5: Input the sub-blocks from Step 4 into the compression chip in sequence, and finally output the compressed bitstream block.

[0039] The following uses 9-band TDICCD image data as an application example to more clearly describe the sub-tile segmentation method based on spaceborne remote sensing image compression of the present invention.

[0040] Taking 9-band TDICCD image data as an application example, in step 2, if the data is in compressed mode, the preprocessed data is split into auxiliary data and image data. For both auxiliary data and image data: each data packet consists of 4 consecutive rows of full chromatographic segments (represented by P-segments) and 8 multispectral segments (represented by B-segments), and is transmitted continuously along the data transmission path according to this data packet. P1~P4 represent the segment identifiers of the 4 consecutive rows of P-segments, and B1~B8 represent the B-segment identifiers. Each data packet is output sequentially in the order of P1~B1~B2~P2~B3~B4~P3~B5~B6~P4~B7~B8. Since the auxiliary data portion is not compressed, it is split into the image data portion and processed separately, as follows... Figure 2 As shown.

[0041] In the 9-band mode, for auxiliary data, auxiliary data packets are extracted and assembled according to the data transmission order, processed uniformly with a fixed number of packets, and directly output as data; specifically: based on Figure 2 The data is processed in units of 128 packets. Auxiliary data from all spectral bands is extracted sequentially to form an auxiliary data package, such as... Figure 3 As shown, the extracted data is directly output.

[0042] For image data, based on the data transmission order and the number of pixels, the size of the compressed sub-tiles is determined, and a row-separated sub-tile segmentation operation is performed. If the remaining number of rows is less than the preset size, padding is performed to ensure the uniformity of the sub-tile sizes; specifically:

[0043] Since the image data needs to be compressed by a compression chip, sub-tile segmentation is required before compression. Schematic diagrams of traditional sub-tile segmentation and the sub-tile segmentation method of this invention are shown below. Figure 4 and Figure 5As shown, the traditional sub-tile segmentation method involves receiving mixed spectral data, temporarily storing it in a buffer, restoring it to complete panchromatic / multispectral spectral data, segmenting the data into sub-tiles, and generating heterogeneous sub-tiles. This method requires first extracting the panchromatic and multispectral spectral data, and then independently performing packet assembly operations on the two types of data. Since the sub-tile size is fixed, the number of packets required for assembling the two types of spectral data differs significantly. For example, to assemble one fixed-size panchromatic sub-tile from the panchromatic spectral data, only 128 transmission packets are needed to form four panchromatic sub-tiles. However, to assemble one fixed-size multispectral sub-tile from the multispectral spectral data, 512 transmission packets are needed. Therefore, a mismatch in the number of transmission packets corresponding to the same sub-tile occurs, leading to asynchronous compression of different spectral data. This is the core reason why the traditional method requires additional caching of a large amount of data and has redundant processes, consuming onboard resources and CPU usage, and also making it inconvenient to subsequently restore the data to the whole-satellite format.

[0044] The sub-block segmentation process in step 4 of this invention involves receiving mixed spectral segment row data, directly segmenting it, and generating sub-blocks with matching row counts for the full chromatographic segment and multispectral segment. This eliminates the need to split the mixed data stream into independent full chromatographic and multispectral segment data. Instead, it directly integrates the row data of both the full chromatographic segment and multispectral segment within the same transmission packet (i.e., data packet), and then assembles sub-blocks of a fixed size based on a unified packet count. For example, to assemble sub-blocks of a fixed size, a total of 512 transmission packets are required, assembling both the row data for the full chromatographic segment and the row data for the multispectral segment sub-blocks simultaneously. This achieves complete uniformity in the packet counts for full chromatographic and multispectral segment sub-blocks, eliminating the need for additional caching of different packet count data for different spectral segments. Different spectral segment data can be compressed synchronously. By omitting redundant segmented data restoration steps, it achieves efficient processing with low cache usage, fundamentally solving the problems of process redundancy and cache burden in traditional methods, saving on-board resources, and facilitating subsequent data restoration to the whole-satellite format.

[0045] This paper presents two strategies for sub-tile segmentation: traditional sub-tile segmentation and the row-separated sub-tile segmentation proposed in this invention. A quantitative analysis is conducted on the images output from the entire compression and decompression process to explore whether different segmentation methods significantly affect image quality, providing data support for the rationality and engineering applicability of the segmentation strategies. The results are as follows: Figure 6 As shown in Table 1:

[0046] Table 1. Comparison of decompression and image restoration metrics for different sub-tile segmentation methods at different compression ratios.

[0047]

[0048] Figure 6The images shown are (a) the original image, (b) the image compressed and decompressed using the traditional sub-tile segmentation method, and (c) the image compressed and decompressed using the row-separation sub-tile segmentation method of this invention. It can be seen that the circular details of the bird's nest, the rectangular outline of the building, and the straight edges of the road are all clearly preserved in the image processed and restored using the sub-tile segmentation method of this invention, without any blurring, block artifacts, or other abnormalities. Furthermore, there is no significant difference in image quality compared to the traditional sub-tile segmentation method.

[0049] Table 1 quantifies the restored images of different sub-tile segmentation methods under different compression ratios based on two core image quality evaluation metrics: Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity (SSIM). Under the three compression ratios, the PSNR difference between the sub-tile segmentation method of this invention and the traditional sub-tile segmentation method is only 5.59 dB, and the SSIM difference is only 0.0715, both below the critical threshold affecting image interpretation. Furthermore, the restored images processed by the sub-tile segmentation method of this invention retain their key structures and boundary features completely, meeting the requirements of subsequent classification, interpretation, monitoring, and other practical applications.

[0050] In summary, the sub-plot segmentation method of the present invention achieves engineering advantages such as no need to split spectral segments, uniform packet count, low cache usage, and short preprocessing latency with minimal image quality loss. It is more suitable for the actual working conditions of spaceborne platforms with limited resources and is an optimized solution that balances image quality reliability and engineering practicality.

[0051] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A sub-tile segmentation method based on spaceborne remote sensing image compression, characterized in that, Includes the following steps: Step 1: Perform preprocessing operations on the received spaceborne remote sensing image data stream, filtering and retaining the valid data portion; Step 2: Determine whether the data after preprocessing in Step 1 is in compression mode. If not, output the corresponding data directly according to the target required spectral band. If so, split the preprocessed data into two parts: auxiliary data and image data. Step 3: For auxiliary data, extract and assemble auxiliary data packets according to the data transmission order, process them uniformly with a fixed number of packets, and directly output the data; Step 4: For image data, based on the data transmission order and the number of pixels, determine the size of the compressed sub-plot and perform row-separated sub-plot segmentation. If the remaining number of rows is less than the preset size, fill the remaining rows to ensure the uniformity of the sub-plot size. Step 5: Input the sub-blocks from Step 4 into the compression chip in sequence, and finally output the compressed bitstream block.

2. The sub-tile segmentation method according to claim 1, characterized in that, Step 1 specifically involves: filtering and retaining valid data portions of the received spaceborne remote sensing image data stream by judging the frame header, frame tail, and byte length.

3. The sub-tile segmentation method according to claim 1, characterized in that, This sub-plot segmentation method is applicable to sub-plot segmentation of 9-band TDICCD image data.

4. The sub-tile segmentation method according to claim 3, characterized in that, In step 4, the image data is composed of a data packet consisting of 4 consecutive rows of full chromatographic segments and 8 spectral segments of multispectral segments, and is continuously transmitted along the data transmission path according to this data packet.

5. The sub-tile segmentation method according to claim 4, characterized in that, Each set of four consecutive full-spectrum segments is labeled P1~P4, and each set of eight multispectral segments is labeled B1~B8. Each data packet is then output in the following order: P1~B1~B2~P2~B3~B4~P3~B5~B6~P4~B7~B8.

6. The sub-tile segmentation method according to claim 4, characterized in that, In step 4: The full chromatographic segment and multispectral segment in each data packet are divided into sub-blocks with a pixel count of 1536. After the full chromatographic segment and multispectral segment data are divided, a total of 24 sub-blocks are formed.

7. The sub-tile segmentation method according to claim 5, characterized in that, Step 3 specifically involves: Using 128 data packets as a processing unit, the auxiliary data of all spectral bands are extracted sequentially according to the data transmission order of P1~B1~B2~P2~B3~B4~P3~B5~B6~P4~B7~B8 to form an auxiliary data packet, which is then directly output after extraction.

Images