Compression coding processing method and system for observation data of low earth orbit satellite

Abstract

The invention discloses a compressed encoding processing method and system for observation data of a low-orbit satellite. The method relates to the technical field of compression coding processing and comprises the following steps of task perception analysis and cooperative compression, task efficiency driving regulation and control, cooperative state monitoring and evaluation and certainty constraint cooperation. According to the method, a target detection result and an ROI mask are generated through on-satellite intelligent analysis, and differential compression of task perception is implemented accordingly. Whether task efficiency regulation is started or not is decided based on the calculated end-to-end task efficiency evaluation value, then the link state is monitored in real time, and the on-satellite cooperative processing certainty is evaluated. Finally, whether collaborative accuracy regulation and control are carried out or not is decided according to the certainty, and a high-reliability compression result is output, so that driving conversion from data fidelity to task utility is realized; the problem that in the prior art, due to cooperative splitting between on-orbit intelligent processing and compressed encoding, compressed encoding processing of low-orbit satellite observation data is low in reliability is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of compression encoding processing, in particular to a compression encoding processing method and system for low-orbit satellite observation data. BACKGROUND

[0002] After the low-orbit satellite collects observation data (including optical images, radar echo signals, etc.) through optical, radar and other loads, it completes data format standardization, noise suppression and regional segmentation. For optical images, SCS-LBP operators combined with sparse coding advantages are used to realize background and foreground segmentation, separate deep space background, star static background area and debris, spacecraft and other dynamic foreground areas, and logarithmic transformation preprocessing is performed on radar echo data. For multi-satellite cooperative observation scenes, SIFT (Scale-Invariant Feature Transform) operators are used to extract image feature points and descriptors, identify overlapping areas of adjacent satellite observation images, and determine the overlapping width and boundary coordinates by using the area equivalent method.

[0003] The mainstream CCSDS (Consultative Committee for Space Data Systems) standard algorithm in the field of satellite is used to perform multi-scale decomposition on the image through 9 / 7 wavelet lifting format discrete wavelet transform, convert pixel information into low-frequency approximation component and high-frequency detail component, and then implement non-uniform quantization on the high-frequency component. Subsequently, the quantization coefficients are scanned by bit plane parallel scanning combined with adaptive arithmetic coding for entropy coding; the number of quantization bits is dynamically adjusted according to the data statistical characteristics, and for scenes that need to retain complete information, JPEG GLS algorithm is selected to complete lossless / near lossless compression through context modeling, prediction error calculation and Golomb-Grice coding. For multi-satellite overlapping area data, aggregate sparse coding technology is used to initialize dictionary tensor and auxiliary tensor, then obtain the dictionary tensor through iterative update and Fourier transform, and aggregate the dictionary tensors of adjacent satellites. During the encoding process, FPGA (Field-Programmable Gate Array) hardware platform is used to ensure real-time performance, block compression strategy is used, pipeline and parallel architecture are used to realize data parallel processing, on-chip RAM (Random Access Memory) is used to cache dictionaries and intermediate data to reduce external storage access delay, and CSD (Canonical Signed Digit) coding is used to replace multiplication operation to reduce resource occupation.

[0004] The encoded coefficient tensor is quantized into integer values, rearranged into importance mapping and level mapping, and subjected to an overlap-aware adaptive binary arithmetic coding algorithm for the overlapping area, a probability model is updated in combination with the pixel context of the current view and the reference view, and a compressed code stream containing a transition tensor and a position tensor is output, and the segmented background region data is also transmitted synchronously, and the code stream is downlinked to the ground station after the payload of the laser communication on the satellite is chained.

[0005] For example, a multi-channel satellite image lossless compression method disclosed in Chinese patent CN103227920B comprises: for multi-channel satellite image data, on the basis of existing lossless compression coding technology, the similarity of satellite cloud images between different channels is fully utilized, the difference of the image is calculated first, and then lossless compression coding is performed.

[0006] For example, a large compression ratio satellite remote sensing image compression method based on a deep auto-encoding network disclosed in Chinese patent CN105163121B comprises: a plurality of auto-encoders are cascaded and stacked to form a deep auto-encoding network; a set of training image data is input to the deep auto-encoding network, the network is trained to obtain optimized network parameters, and a deep compression network and a deep decompression network are obtained; a remote sensing image to be compressed is input to the deep compression network, high-order sparse features are obtained, the features are quantized and coded to obtain a final compressed code stream; the received code stream is dequantized and coded, high-order sparse features are obtained and input to the deep decompression network, and the final output of the network is a decompressed remote sensing image.

[0007] However, in the process of implementing the technical scheme of the embodiments of the present application, the applicant finds that the above-mentioned technology at least has the following technical problems: In the prior art, the traditional on-board data compression technology is rooted in the classic Shannon rate-distortion theory, and the optimization goal is to minimize the overall mathematical error between the reconstructed image and the original image at the pixel level under a given code rate. This paradigm regards images as a collection of indistinguishable signals, and is committed to preserving all pixel information, while completely ignoring the semantic information extraction requirements of subsequent intelligent analysis tasks (such as target detection and feature classification). The background pixels (such as large areas of uniform ocean, clouds and deserts) with extremely low or zero task value are encoded and transmitted with high precision, while the real key target information is submerged in the redundant data and transmitted together, resulting in a very low end-to-end efficiency of the system from the perspective of information utility, as most of the downlink bandwidth is occupied by invalid data. In engineering implementation, the compression module adopts a pre-solidified and globally unified coding strategy (such as fixed quantization parameters), which cannot dynamically and differentially allocate code rate resources according to the semantic importance (such as detected target regions and confidence of regions of interest) perceived by the image content or analysis module in real time, further exacerbating the processing delay and limiting the flexibility of the system in dealing with multiple tasks and dynamic scenarios. Under the existing serial architecture, high-complexity intelligent analysis models and high-performance compression encoders need to independently occupy the already scarce on-board computing and storage resources, and their simple concatenation increases the end-to-end processing delay. More importantly, the "black box" characteristics of intelligent analysis models (especially deep learning models) and their performance uncertainty are inherently contradictory to the stringent requirements of high reliability and determinism of space systems, greatly increasing the design, verification and on-orbit operation risk of deep collaboration between the two, and there is a problem of low reliability of compression encoding processing of low-orbit satellite observation data due to the collaborative fragmentation between on-orbit intelligent processing and compression encoding. SUMMARY

[0008] In order to solve the technical problems of the prior art that the compression encoding processing reliability of low-orbit satellite observation data is low due to the collaborative fragmentation between on-orbit intelligent processing and compression encoding, the embodiments of the present application provide a compression encoding processing method and system for low-orbit satellite observation data. The technical solution is as follows: On the one hand, a compression encoding processing method for low-orbit satellite observation data is provided, which comprises: S1, performing intelligent analysis of observation data on-orbit to generate target detection results and region of interest masks containing semantic information, implementing task-aware differential compression encoding on original observation data based on the target detection results and the region of interest masks, and obtaining task efficiency data for quantifying the efficiency of converting low-orbit satellite observation data into effective information from the start of obtaining original observation data from the satellite to the final generation of the process on the ground to meet the preset application requirements, to obtain an end-to-end task efficiency evaluation value; S2, determining whether to perform task efficiency intelligent regulation according to the end-to-end task efficiency evaluation value, realizing the change from data fidelity driving to task utility driving, thereby realizing the collaborative optimization of information processing efficiency and task success rate; if yes, then monitoring the bandwidth occupancy rate of the downlink and the delay state of the on-board processing pipeline in real time after regulation; if no, then directly monitoring the bandwidth occupancy rate of the downlink and the delay state of the on-board processing pipeline in real time; S3, obtaining collaborative processing data for quantifying the collaborative accuracy of on-orbit intelligent analysis and compression encoding of low-orbit satellites on-orbit during real-time monitoring of the bandwidth occupancy rate of the downlink and the delay state of the on-board processing pipeline, to obtain on-board collaborative processing certainty; S4, determining whether to perform collaborative processing accuracy intelligent regulation according to the on-board collaborative processing certainty, thereby maintaining the reliability of the on-board processing link and the credibility of the output results; if yes, then outputting the compression encoding processing result after regulation; if no, then directly outputting the compression encoding processing result.

[0009] On the other hand, a compression coding and processing system for low-Earth orbit (LEO) satellite observation data is provided. This system includes: a mission-aware analysis and collaborative compression module, a mission performance-driven control module, a collaborative status monitoring and evaluation module, and a deterministic constraint-based collaborative control module. Specifically, the mission-aware analysis and collaborative compression module performs intelligent analysis of observation data in orbit, generating target detection results and region-of-interest (ROI) masks containing semantic information. Based on these results, it performs mission-aware differentiated compression coding on the raw observation data to obtain mission performance data. This data quantifies the efficiency of converting LEO satellite observation data into effective information from the initial acquisition of raw observation data from the satellite to the final generation of preset application requirements on the ground, yielding an end-to-end mission performance evaluation value. The mission performance-driven control module determines whether to execute intelligent mission performance control based on the end-to-end mission performance evaluation value, realizing a shift from data fidelity-driven to mission-constrained control. The shift towards service utility-driven transformation achieves optimal synergy between information processing efficiency and task success rate. If executed, the downlink bandwidth utilization and on-board processing pipeline latency status are monitored in real time after adjustment; otherwise, the downlink bandwidth utilization and on-board processing pipeline latency status are directly monitored in real time. The collaborative status monitoring and evaluation module acquires collaborative processing data during real-time monitoring of downlink bandwidth utilization and on-board processing pipeline latency status. This data is used to quantify the collaborative accuracy of on-orbit intelligent analysis and compression coding collaboration on the low-Earth orbit satellite, obtaining on-board collaborative processing determinism. The deterministic constraint collaborative adjustment module determines whether to perform intelligent adjustment of collaborative processing accuracy based on the on-board collaborative processing determinism, thereby maintaining the reliability of the on-board processing link and the credibility of the output results. If executed, the compressed coding processing result is output after adjustment; otherwise, the compressed coding processing result is directly output.

[0010] Beneficial effects The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: 1. By intelligently analyzing observation data in orbit, target detection results and region-of-interest (ROI) masks containing semantic information are generated. Based on the target detection results and ROI masks, task-aware differentiated compression coding is performed on the raw observation data to obtain task performance data. This data is used to quantify the efficiency of converting low-Earth orbit (LEO) satellite observation data into effective information from the initial acquisition of raw observation data from the satellite to the final generation of preset application requirements (such as target identification reports and disaster assessment maps) on the ground. An end-to-end task performance evaluation value is obtained. Based on the end-to-end task performance evaluation value, it is determined whether to perform intelligent task performance adjustment. This realizes the transformation from data fidelity-driven to task utility-driven, thereby achieving the optimal synergy between information processing efficiency and task success rate. In the process of real-time monitoring of downlink bandwidth utilization and on-board processing pipeline latency, collaborative processing data is obtained. This data is used to quantify the collaborative accuracy of on-orbit intelligent analysis and compression coding on the LEO satellite, obtaining on-board collaborative processing determinism. Based on the on-board collaborative processing determinism, it is determined whether to perform intelligent adjustment of collaborative processing accuracy, thereby maintaining the reliability of the on-board processing link and the credibility of the output results, and thus improving the reliability of compression coding processing.

[0011] 2. By implementing dynamic window scheduling for space-ground collaboration oriented towards mission timeliness, ground processing latency is incorporated into the real-time decision-making closed loop of the onboard transmission window. Through dynamic compression or adaptation of transmission time slots, it ensures that even under tight end-to-end time constraints, the space-ground system can still prioritize the completion of the entire process from acquisition to application of high-value data within the mission timeliness threshold. This transforms limited transit window resources into deterministic information delivery capabilities. Through adaptive background compression based on channel congestion awareness, channel state (bandwidth occupancy) is used as a direct control variable for the compression intensity of the background area. By establishing an adaptive mapping relationship between occupancy and compression intensity, the potential value of background information is preserved when the channel is idle, and extreme compression or even selective discarding is performed when the channel is congested. This intelligently releases downlink bandwidth pressure while maintaining the quality of critical foreground information, achieving dynamic optimal allocation of communication link resources between valuable information and redundant background data, thereby improving the reliability of compression coding processing.

[0012] 3. Through adaptive queue regulation based on processing delay awareness, the state of the ROI mask generation delay, a key bottleneck in the processing pipeline, is taken as a forward-looking indicator for queue management. By establishing a mapping relationship between delay and overflow rate, the water level line is dynamically adjusted and the discard strategy is switched, realizing the transition from "passive response to overflow" to "active prevention of backlog". Thus, when the intelligent processing delay occurs, the buffer queue can be reorganized in advance, the high-value data flow is prioritized, and the continuous smoothness and overall stability of the data processing link are ensured. Through adaptive trigger noise margin regulation based on jitter awareness, the real-time jitter statistics of the cooperative trigger delay and the system required cooperative processing determinacy are taken as joint inputs, and the noise margin threshold of trigger judgment is dynamically improved, effectively filtering out false trigger signals caused by spatial environmental interference or system load fluctuations, suppressing the inter-module cooperative out-of-order caused by false triggers, ensuring the timing accuracy and reliability of the interactive action between intelligent analysis and compression encoding, and thus improving the compression encoding processing reliability. BRIEF DESCRIPTION OF DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0014] Figure 1 A flow chart of the compression encoding processing method of low-orbit satellite observation data provided by the embodiment of the present application; Figure 2 An adaptive background compression flow chart of the compression encoding processing method of low-orbit satellite observation data provided by the embodiment of the present application based on channel congestion awareness; Figure 3 An adaptive queue regulation flow chart of the compression encoding processing method of low-orbit satellite observation data provided by the embodiment of the present application based on processing delay awareness; Figure 4 A structural schematic diagram of the compression encoding processing system of low-orbit satellite observation data provided by the embodiment of the present application. DETAILED DESCRIPTION

[0015] The technical solutions provided by the present application will be described below with reference to the drawings.

[0016] In order to facilitate the understanding of the embodiments of the present application, the following points will be explained first: First, in the present application, "at least one" means one or more, and "multiple" means two or more. The "and / or" describes the association between the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the following three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after it, but does not rule out the case where the associated objects before and after it represent an "and" relationship. The specific meaning can be understood in conjunction with the context. "At least one" or similar expressions refer to any combination of these items, including single or multiple items in any combination. For example, at least one of a, b, or c can represent: a, b, c; a and b; a and c; b and c; or a and b and c. Where a, b, and c can be singular or plural.

[0017] Second, in the present application, the use of prefixes such as "first", "second", etc. is only for the convenience of distinguishing and describing different things belonging to the same name category, and does not constrain the order, size or quantity of the things. For example, "first message" and "second message" are only different messages, and there is no time sequence, size relationship or priority relationship between them.

[0018] In the embodiments of the present application, sometimes the subscript such as W1 may be written in the form of non-subscript such as W1, and when the distinction is not emphasized, the meaning expressed is consistent.

[0019] In order to make the technical problems, technical solutions and advantages to be solved by the present application more clear, the following will be described in detail in conjunction with the drawings and specific embodiments.

[0020] As shown in Figure 1 The method provided by the embodiment of the present application includes the following steps: S1, intelligent analysis of observation data in orbit, generating target detection results and region of interest masks containing semantic information, based on the target detection results and the region of interest masks, implementing task-aware differential compression encoding on the original observation data, obtaining task performance data, which is used to quantify the efficiency of converting low-orbit satellite observation data into effective information from the beginning of obtaining the original observation data from the satellite to the end of generating a preset application requirement such as a target identification report or a disaster assessment map on the ground, and obtaining an end-to-end task performance evaluation value.

[0021] It should be understood that the specific steps for obtaining the end-to-end task performance evaluation value are as follows: The task performance data includes low-orbit satellite observation data downlink transmission delay, end-to-end information utility bandwidth ratio and low-orbit satellite observation data compression delay. The low-orbit satellite observation data downlink transmission delay refers to the total time consumption from the output end of the satellite payload to the ground receiving station data demodulation completion end through the satellite-ground / inter-satellite downlink. The ground receiving station extracts the time stamp after demodulation to calculate the difference between the ground receiving time and the satellite transmission time, that is, the low-orbit satellite observation data downlink transmission delay. The end-to-end information utility bandwidth ratio refers to the actual available band of the downlink collected by the spectrum analyzer of the ground receiving station, and the effective data transmission bandwidth is obtained by deducting the signaling and error correction overhead. The ratio of the effective data transmission bandwidth to the total downlink data transmission bandwidth is the end-to-end information utility bandwidth ratio. The low-orbit satellite observation data compression delay refers to the total calculation time consumption from entering the compression to outputting the compressed data frame. The input / output end of the on-board compression module is embedded with a time stamp, and the time stamp data is transmitted to the ground through the on-board telemetry system. The difference between the two time stamps is the low-orbit satellite observation data compression delay.

[0022] The transmission delay influence component is obtained by combining the ratio of the transmission delay critical value to the low-orbit satellite observation data downlink transmission delay with the transmission delay correction factor. The bandwidth ratio influence component is obtained by combining the ratio of the bandwidth ratio critical value to the end-to-end information utility bandwidth ratio with the bandwidth ratio correction factor. The compression delay influence component is obtained by combining the ratio of the compression delay critical value to the low-orbit satellite observation data compression delay with the compression delay correction factor. The end-to-end task performance evaluation value is obtained by coupling the transmission delay influence component, the bandwidth ratio influence component and the compression delay influence component. The specific limit expression of the end-to-end task performance evaluation value is: ; ; In the formula, R represents the end-to-end task performance evaluation value; s1 represents the transmission delay correction factor obtained from the compression encoding database; s2 represents the bandwidth ratio correction factor obtained from the compression encoding database; s3 represents the compression delay correction factor obtained from the compression encoding database; N0 represents the transmission delay critical value obtained from the compression encoding database; E0 represents the bandwidth ratio critical value obtained from the compression encoding database; F0 represents the compression delay critical value obtained from the compression encoding database; N represents the low-orbit satellite observation data downlink transmission delay; E represents the end-to-end information utility bandwidth ratio; and F represents the low-orbit satellite observation data compression delay.

[0023] It should be understood that the longer the downlink transmission latency of low-Earth orbit (LEO) satellite observation data, the lower the number of effective frames / resolution / update frequency that can be transmitted per unit time, and the lower the end-to-end information utility bandwidth ratio. The longer the downlink transmission latency of LEO satellite observation data, the more severe the data throughput capacity of the satellite-to-ground channel is, resulting in serious data backlog. This "forces" the compression latency of LEO satellite observation data to be shortened. The shorter the compression latency of LEO satellite observation data, the more complex and refined the intelligent processing and analysis being performed on the satellite, leading to a significant improvement in the end-to-end information utility bandwidth ratio. Meanwhile, there is a negative correlation between the downlink transmission latency of low-Earth orbit (LEO) satellite observation data and the end-to-end mission performance evaluation value. The longer the downlink transmission latency of LEO satellite observation data, the more incomplete key frames are transmitted, the more time-sensitive the data becomes, and the less effective the data is, resulting in a lower end-to-end mission performance evaluation value. Conversely, there is a positive correlation between the end-to-end information utility bandwidth ratio and the end-to-end mission performance evaluation value. The higher the end-to-end information utility bandwidth ratio, the more effective information can be transmitted with the same bandwidth (e.g., higher resolution, more frames, shorter update cycles), thus improving both the accuracy and timeliness of the mission, resulting in a higher end-to-end mission performance evaluation value. Furthermore, there is a positive correlation between the compression latency of LEO satellite observation data and the end-to-end mission performance evaluation value. The longer the compression latency of LEO satellite observation data, the more complex and higher compression ratio compression strategy is used on-board / ground, and this strategy does not cause data distortion or exceed the mission timeliness threshold, ultimately improving the end-to-end mission performance, resulting in a higher end-to-end mission performance evaluation value.

[0024] S2 determines whether to perform intelligent task performance control based on end-to-end task performance evaluation values, realizing the transformation from data fidelity-driven to task utility-driven, thereby achieving optimal synergy between information processing efficiency and task success rate. If executed, the downlink bandwidth utilization and on-board processing pipeline latency status are monitored in real time after the control is implemented. If not, the downlink bandwidth utilization and on-board processing pipeline latency status are monitored directly in real time.

[0025] Furthermore, the specific steps for determining whether to perform intelligent task performance control are as follows: if the end-to-end task performance evaluation value is greater than or equal to the task performance benchmark value, then intelligent task performance control is not performed; otherwise, dynamic window scheduling based on satellite-ground coordination and adaptive background compression based on channel congestion awareness are performed to meet task timeliness.

[0026] It should be further explained that the specific steps for performing mission-time-oriented space-ground collaborative dynamic window scheduling are as follows: For each low-Earth orbit satellite observation mission, a maximum end-to-end allowable delay threshold is predefined. The maximum end-to-end allowable delay threshold represents the maximum total time allowed from the time of low-Earth orbit satellite observation data acquisition to the time of final delivery to the end user.

[0027] Before and during each satellite pass, the current ground processing and distribution delay is monitored and obtained in real time, and according to the end-to-end maximum allowed delay threshold and the ground processing and distribution delay, a total time window available for star-ground collaborative use for on-board processing and downlink transmission is obtained, which represents the difference between the end-to-end maximum allowed delay threshold and the ground processing and distribution delay.

[0028] The task-time-oriented star-ground collaborative dynamic window scheduling is also performed, and further includes: Based on the end-to-end task performance evaluation value and the total time window, in combination with the preset time window of the satellite pass, the following logic is used to dynamically determine and instruct the satellite to perform the adjusted actual scheduling transmission time window, specifically: If the total time window is greater than or equal to the preset time window, it indicates that the ground processing margin is sufficient, and the satellite can be instructed to perform low-orbit satellite observation data transmission in the entire pass preset time window, that is, the actual scheduling transmission time window is the preset time window.

[0029] If the total time window is less than the preset time window, it indicates that the total time is tight, and the transmission window is compressed to reserve sufficient time for ground processing. Based on the end-to-end task performance evaluation value and the current total time window, the actual scheduling transmission time window is output by inputting into the predefined distribution delay-transmission time window mapping relationship, and the satellite is instructed to perform high-priority data priority transmission only in the actual scheduling transmission time window, wherein the high-priority data refers to the data unit marked by the system as having the highest information value and time urgency for completing the preset observation task in the process of intelligent processing and collaborative compression on the satellite, and thus being given the priority of transmission.

[0030] The start time, end time and corresponding data transmission priority strategy of the obtained actual scheduling transmission time window are encapsulated as scheduling instructions, so that the low-orbit satellite observation data transmission is performed according to the adjusted actual scheduling transmission time window when the satellite enters the previous ground station visible window.

[0031] In the embodiment, firstly, an end-to-end maximum allowable delay threshold is predefined for each observation task, which establishes a global timeliness constraint from data acquisition to user reception. Before and during each satellite pass, the current ground processing and distribution delay is monitored and obtained in real time, and then the ground delay is deducted from the maximum allowable delay to obtain a total time window dedicated to on-board processing and downlink transmission, which incorporates the variable processing delay of the ground system into the real-time planning of satellite transmission, breaking the timeliness disconnection problem caused by traditional independent planning of satellite and ground. Then, based on the calculated total time window and the theoretical visible window (preset time window) of the satellite pass, combined with the real-time evaluated end-to-end task performance value, dynamic decision is made through the predefined distribution delay-transmission time window mapping relationship: if the total time window is sufficient (greater than or equal to the preset window), the satellite is instructed to transmit in the entire pass window with the most stable parameters, fully utilizing the channel resources to maximize data throughput; if the total time window is tight (less than the preset window), it means that the transmission time must be compressed to reserve a margin for ground processing, at this time the system calculates the compressed actual scheduling transmission time window according to the task performance value and the tight time window through the mapping relationship, and instructs the satellite to only transmit high-priority data (i.e. data units marked by on-board intelligent processing with the highest information value and timeliness urgency) in the compressed window, this mechanism ensures that in the critical state where the end-to-end time limit is about to be broken, the system can automatically focus on the timely delivery of core task information, realizing the leap from "best effort transmission" to "quality and on-time" delivery. Finally, the calculated actual window parameters and priority strategy are packaged into scheduling instructions, which are uplinked when the satellite enters the visible window of the previous ground station, forming a closed-loop control of satellite-ground cooperation, so that the satellite's transmission behavior can adapt to the changes in ground processing state in a forward-looking manner, and finally the limited pass time is converted into deterministic task timeliness guarantee at the system level.

[0032] It should be noted that, as Figure 2As shown, the adaptive background compression flowchart based on channel congestion awareness of the low-orbit satellite observation data compression and encoding processing method provided by the embodiment of the application has the specific flow as follows: first, the real-time monitoring and evaluation link is started from the "start monitoring", and the downlink total bandwidth occupancy is continuously collected; then, through the "bandwidth occupancy rate judgment" node, the real-time occupancy rate is compared with the preset threshold value, if the occupancy rate is less than or equal to the threshold value, it is determined that the link is in the "channel idle state", at this time, the compression strength of the current background area is directly maintained, the current regulation and control is completed, and the "end regulation and control" is entered; if the occupancy rate is greater than the threshold value, it is determined that the link is in the "channel congestion state", and then the joint decision input link is entered, the "downlink total bandwidth occupancy" and "end-to-end task performance evaluation value" two core parameters are integrated, the "occupancy rate-maximum compression strength mapping relationship" is combined, the maximum compression strength of the target background area is calculated and output, and finally the maximum compression of the background area is applied to release the bandwidth resource, and after the regulation and control is completed, the "end regulation and control" is also entered. The whole process dynamically adjusts the compression strategy of the background area by distinguishing the channel state, guarantees the data quality in the idle state, releases the bandwidth in the congestion state, and realizes the balance between the bandwidth utilization efficiency and the task performance.

[0033] It needs to be further explained that the specific steps of the adaptive background compression based on channel congestion awareness are as follows: The total bandwidth occupancy of the downlink is monitored in real time, and according to the real-time monitored downlink total bandwidth occupancy and the end-to-end task performance evaluation value, the maximum allowed compression strength of the background area is dynamically adjusted, specifically: if the downlink total bandwidth occupancy is less than or equal to the bandwidth occupancy rate threshold value, it is determined that the channel idle degree is high, the downlink total bandwidth resource is sufficient, and the maximum allowed compression strength of the current background area is maintained.

[0034] If the downlink total bandwidth occupancy is greater than the bandwidth occupancy rate threshold value, it is determined that the channel has reached the congestion state, the downlink total bandwidth occupancy and the end-to-end task performance evaluation value are input into the predefined occupancy rate-maximum compression strength mapping relationship, the maximum compression strength of the target background area is output, the maximum compression of the background area is applied to reduce the data amount, and the bandwidth is reserved for the key information.

[0035] In the embodiment, firstly, the total bandwidth occupancy of the downlink is continuously monitored in real time, which comprehensively reflects the overall consumption of the channel by the downlink data flow of the satellite and all satellites sharing the same ground station resources. Through global bandwidth occupancy awareness, the system can accurately judge the real-time tension degree of the channel resources and provide accurate basis for dynamic resource allocation. When the total downlink bandwidth occupancy is less than or equal to the preset bandwidth occupancy critical value, the system determines that the current channel is high in idle degree and the bandwidth resources are abundant. At this time, the system maintains the current compression strength of the background area, which aims to moderately reserve the potential value of the background information (such as background change analysis for long-term environmental monitoring) when the channel resources are relaxed, avoid the loss of future application potential due to excessive compression, and reflect the strategy of resource flexible management. When the total downlink bandwidth occupancy is greater than the critical value, the system determines that the channel has entered a congested state. At this time, instead of simply and singlely enhancing compression, the system inputs the total downlink bandwidth occupancy and the end-to-end task performance evaluation value reflecting the global task completion efficiency into the pre-defined occupancy-maximum compression strength mapping relationship for joint decision-making. According to the channel congestion degree and the sensitivity of the current task performance, the mapping relationship intelligently calculates and outputs an optimal target background area maximum compression strength. Subsequently, the system applies a corresponding level of compression, or even maximum compression, to the background area according to the strength, directly and intelligently converts the congestion state of the channel into a "squeezing" instruction for redundant background data, thereby rapidly and significantly reducing the volume of invalid data under the premise of maintaining the absolute priority of the quality of key foreground information, actively "clearing obstacles" and "making way" for the transmission of high-value information, realizing resource rebalancing in the bandwidth shortage crisis, and ensuring that the core task performance is still given priority in the channel congestion.

[0036] S3, in the process of monitoring the bandwidth occupancy of the downlink and the time delay state of the on-board processing pipeline in real time, cooperative processing data is obtained, which is used to quantify the cooperative accuracy of the on-board intelligent analysis and compression encoding of the low-orbit satellite, and the on-board cooperative processing certainty is obtained.

[0037] It should be explained that the specific steps for obtaining the on-board cooperative processing certainty are as follows: The cooperative processing data includes a task performance effective value, a ROI content discrimination time delay, and an over-the-top window cooperative completion rate. The task performance effective value refers to that if the task performance intelligent regulation is performed, the re-acquired end-to-end task performance evaluation value is recorded as the task performance effective value, and if not, the current end-to-end task performance evaluation value is recorded as the task performance effective value. The ROI content discrimination time delay refers to the total time consumption from the reception of the first byte of the original observation data frame by the intelligent analysis module on the low-orbit satellite to the output of the mask corresponding to the frame. The difference between the two time stamps is the ROI content discrimination time delay. The ROI content discrimination time delay is calculated by embedding high-precision time stamps in the data input port and the ROI output port of the intelligent analysis module on the satellite, and downloading the time stamp data to the ground station with the telemetry frame. The ROI is a local area in the original observation data frame that has core value for a specific task and needs to be analyzed and reserved. These areas are the carriers of the task target and are the priority objects of subsequent compression encoding and downlink transmission. The over-the-top window cooperative completion rate refers to the number of tasks that successfully complete the whole cooperative process of original data acquisition-ROI content discrimination-compression encoding-downlink transmission within a single over-the-top window of the low-orbit satellite to the ground target, accounting for the proportion of the total number of tasks in the window period. The total number of tasks is calculated by extracting the task number, acquisition time, and data volume in the window period from the telemetry log of the satellite payload, and the ratio of the total number of tasks to the total number of tasks in the window period is recorded as the over-the-top window cooperative completion rate.

[0038] The performance effective correction factor is combined with the ratio of the task performance effective value to the performance effective critical value to obtain a performance effective influence component. The discrimination time delay correction factor is combined with the ratio of the discrimination time delay critical value to the ROI content discrimination time delay to obtain a discrimination time delay influence component. The cooperative completion rate correction factor is combined with the ratio of the over-the-top window cooperative completion rate to the cooperative completion rate critical value to obtain a cooperative completion rate influence component. The performance effective influence component, the discrimination time delay influence component, and the cooperative completion rate influence component are coupled to obtain the on-satellite cooperative processing determinacy. The specific limit expression of the on-satellite cooperative processing determinacy is: ; ; In the formula, Q represents the on-satellite cooperative processing determinacy, v1 represents the performance effective correction factor obtained from the compression encoding database, v2 represents the discrimination time delay correction factor obtained from the compression encoding database, v3 represents the cooperative completion rate correction factor obtained from the compression encoding database, D0 represents the performance effective critical value obtained from the compression encoding database, T0 represents the discrimination time delay critical value obtained from the compression encoding database, P0 represents the cooperative completion rate critical value obtained from the compression encoding database, D represents the task performance effective value, T represents the ROI content discrimination time delay, and P represents the over-the-top window cooperative completion rate.

[0039] It should be understood that the higher the over-the-top window collaborative completion rate is, the more tasks are completed in the window period, the more effective data is obtained on the ground, the higher the data update frequency is, and the higher the task performance effective value is; the longer the ROI content discrimination delay is, the single task processing time exceeds the window period threshold, and a large number of tasks overflow due to the inability to be compressed and downloaded in time, the lower the over-the-top window collaborative completion rate is; which leads to a significant decrease in ROI discrimination accuracy (such as target missed detection, mask misplacement), and the lower the task performance effective value is. At the same time, the task performance effective value and the on-board collaborative processing certainty have a positive correlation, the higher the task performance effective value is, the stronger the alignment accuracy of intelligent analysis and compression, the stronger the process stability, the more controllable the output quality and quantity of effective data, and the higher the on-board collaborative processing certainty is; the ROI content discrimination delay and the on-board collaborative processing certainty have a negative correlation, the longer the ROI content discrimination delay is, the more chaotic the collaborative timing is, and the lower the on-board collaborative processing certainty is; the over-the-top window collaborative completion rate and the on-board collaborative processing certainty have a positive correlation, the higher the over-the-top window collaborative completion rate is, the more stable the single task processing delay is, the more controllable the collaborative triggering timing is, the lower the queue overflow rate is, and the higher the on-board collaborative processing certainty is.

[0040] S4, judging whether to perform collaborative processing accuracy intelligent regulation according to the on-board collaborative processing certainty, thereby maintaining the reliability of the on-board processing link and the credibility of the output result, if yes, outputting the compression encoding processing result after regulation, and if no, directly outputting the compression encoding processing result.

[0041] Further, the specific steps of judging whether to perform collaborative processing accuracy intelligent regulation are: if the on-board collaborative processing certainty is greater than or equal to the collaborative processing reference value, no collaborative processing accuracy intelligent regulation is performed, otherwise, adaptive queue regulation based on processing delay perception and adaptive trigger noise tolerance regulation based on jitter perception are performed.

[0042] It should be noted that, as Figure 3As shown, the low-orbit satellite observation data compression encoding processing method provided by the embodiment of the application is based on the adaptive queue regulation flowchart of the processing time delay perception, and the specific process is as follows: first, the real-time monitoring and evaluation link is started from the “start monitoring”, and the “current ROI mask generation time delay” is continuously collected; then, through the “time delay state judgment” node, the real-time time delay is compared with the preset reference value, if the time delay is less than or equal to the reference value, it is determined that the “intelligent processing is running normally”, at this time, the current queue management strategy and the standard overflow rate threshold value are maintained, the queue is allowed to transmit fluctuation to the flat downstream, the regulation is completed and the “end adjustment” is entered. If the time delay is greater than the reference value, it is determined that the “intelligent processing is delayed”, and then the joint decision input link is entered, the “ROI mask generation time delay” and the “onboard cooperative processing certainty” two core parameters are integrated, the “generation time delay-overflow rate mapping relationship” is combined, and the target queue management and overflow rate threshold value are output; subsequently, the “data loss strategy is switched by lowering the overflow waterproof level”, and at the same time, the “data flow is triggered in advance to reserve the buffer space of data backlog”, so as to relieve the risk of queue congestion caused by processing delay, and after the regulation is completed, it also enters the “end adjustment”. Through the dynamic monitoring of the state of the ROI mask generation time delay, the queue management strategy is adjusted in combination with the cooperative processing certainty, which not only guarantees the flexibility of data transmission when the processing is normal, but also actively optimizes the queue buffer and data scheduling when the delay occurs, effectively maintaining the stability of the onboard intelligent processing link and the continuity of data processing.

[0043] As a further specific description, the specific steps of performing adaptive queue regulation based on processing time delay perception are as follows: In the onboard intelligent processing pipeline, the ROI mask generation time delay of the current frame is monitored in real time, and according to the real-time monitored ROI mask generation time delay and the onboard cooperative processing certainty, the overflow prevention water level line and the data loss strategy of the queue are dynamically adjusted, specifically as follows: If the ROI mask generation time delay is less than or equal to the generation time delay reference value, it is determined that the intelligent processing is running normally, the current onboard buffer queue management and overflow rate are set to the onboard buffer queue management and overflow rate threshold value, so as to allow the data stored in the queue to transmit smoothly downstream.

[0044] If the ROI mask generation time delay is greater than the generation time delay reference value, the intelligent processing is delayed, based on the ROI mask generation time delay and the onboard cooperative processing certainty, the target onboard buffer queue management and overflow rate are output by inputting into the predefined generation time delay-overflow rate mapping relationship, and the onboard buffer queue management and overflow rate are lowered, so as to trigger the data flow in advance, and reserve the buffer space of the low-orbit satellite observation data backlog that may be generated due to the processing speed reduction in the future.

[0045] In this embodiment, in the intelligent processing pipeline on the satellite, the ROI mask generation delay of the current frame is monitored in real time, which directly reflects the processing speed of the intelligent analysis module and is a key indicator for judging whether the data production rate is normal. Through continuous monitoring of the key node delay, the system can prospectively perceive the risk of throughput changes that the entire processing pipeline will face. When the ROI mask generation delay is less than or equal to the preset generation delay reference value, the system determines that the intelligent processing module is running normally, and the data production rate is stable. At this time, the overflow prevention water level line and management strategy of the on-board buffer queue are maintained at a preset higher threshold, which allows the queue to accumulate more data to effectively absorb and buffer the short-term rate fluctuations that may occur in the downstream compression, encoding or transmission link, thereby improving the overall smoothness and throughput efficiency of the system and avoiding frequent data flow interruptions caused by minor instabilities in the downstream. When the ROI mask generation delay is greater than the reference value, the system determines that the intelligent processing has been delayed, indicating that the data production rate will decrease, and if not intervened, the queue will be gradually emptied in the follow-up, thereby causing the processing pipeline to be interrupted; more seriously, if the downlink transmission rate does not change at this time, the emptied queue cannot provide enough data, which will cause the valuable downlink bandwidth to be idle. To cope with this risk, the system does not passively wait, but inputs the "ROI mask generation delay" and "on-board cooperative processing determinacy" representing the system's cooperative stability into the pre-defined generation delay-overflow rate mapping relationship. Based on the severity of the delay and its impact on system determinacy, the mapping relationship intelligently calculates the target queue management and overflow rate threshold to which it should be adjusted down. The system then adjusts the overflow prevention water level line accordingly and switches to a more aggressive data discarding or preferential flow strategy that matches it. The queue management mechanism is transformed from passive "overflow reaction" to active "prevention before speed reduction". By triggering data flow or selectively discarding low-value data in advance before the water level line hits the bottom, the system actively "reserves" buffer space for the inevitable data backlog that will be generated due to the continuous slowdown of processing, while ensuring that high-priority data can be processed first. This is like stocking "data food" for the upcoming "data low tide period", thereby avoiding pipeline interruption and bandwidth idling. When the intelligent processing performance fluctuates, the system maximizes the continuity and overall throughput efficiency of the data processing link, ensuring the stable operation of the system in non-ideal conditions.

[0046] As a further specific description, the specific steps of the adaptive trigger noise tolerance regulation based on jitter perception are as follows: A predefined cooperative trigger delay jitter reference value is defined, and the reference value of the cooperative trigger delay jitter is collected and recorded in the star cooperative processing process. The actual delay from the analysis completion to the compression trigger is collected and recorded, and the current jitter statistical quantity of the actual delay sequence in the preset window is calculated in real time. According to the star cooperative processing certainty and the current jitter statistical quantity, the noise margin threshold for determining the analysis completion signal is dynamically adjusted, and the specific steps are as follows: If the current jitter statistical quantity is less than or equal to the cooperative trigger delay jitter reference value, it is determined that the cooperative trigger timing is stable, and the current trigger threshold noise margin threshold is maintained.

[0047] The adaptive trigger noise margin regulation based on jitter perception further comprises: If the current jitter statistical quantity is greater than the cooperative trigger delay jitter reference value, it is determined that the cooperative trigger timing is abnormally jittered, and the trigger threshold noise margin threshold is increased to suppress the early or delayed false trigger events caused by noise. Specifically, based on the star cooperative processing certainty and the current jitter statistical quantity, an input is input into the predefined delay jitter-marginal threshold mapping relationship, an output adjustment gain coefficient is output, the adjustment gain coefficient is combined with the trigger threshold noise margin threshold, and a target trigger threshold noise margin threshold is obtained. The threshold is combined to improve the threshold of trigger determination, thereby ensuring the accuracy and certainty of the cooperative action in timing.

[0048] In this embodiment, first, a cooperative trigger delay jitter reference value is predefined, which represents the allowed timing fluctuation range of the key cooperative action of the intelligent analysis module completing processing to successfully triggering the compression module under the normal state of system calibration. In the process of on-board cooperative processing, the system continuously collects and records the actual time delay of each "analysis completion" to "compression trigger" with high precision, and based on the time delay sequence within a sliding time window, the standard deviation or mean absolute difference is calculated as the current jitter statistic. This continuous monitoring mechanism provides a direct means for the system to quantitatively evaluate the timing stability of the cooperative link, enabling it to detect implicit interference caused by single event effects, power noise or software scheduling uncertainty. When the calculated current jitter statistic is less than or equal to the preset reference value, the system determines that the cooperative trigger timing is in a stable state, and the existing noise margin threshold is maintained unchanged. This avoids unnecessary adjustments when the system is running smoothly, reduces the additional uncertainty that may be introduced by frequent changes in the threshold, and ensures the optimal performance of the system under normal conditions. When the current jitter statistic is greater than the reference value, the system determines that the cooperative trigger timing has abnormal jitter, which means that the interference level is rising and the risk of false triggering (advance or delay) is significantly increased. At this time, instead of simply increasing the threshold by a fixed step, the system inputs the "current jitter statistic" and the "on-board cooperative processing certainty" reflecting the overall cooperative reliability of the system into a predefined time delay jitter-margin threshold mapping relationship. According to the severity of the jitter (interference intensity) and the current requirement for certainty (reliability tolerance) of the system, the mapping relationship intelligently decides and outputs a "regulation gain coefficient". Subsequently, the system combines this gain coefficient with the current noise margin threshold (e.g. multiplication or addition) to generate a new, higher "target noise margin threshold". This achieves "intelligent noise tolerance", which effectively filters out false trigger signals or glitches that are above the original threshold but below the new threshold by raising the signal threshold for trigger decision, thereby fundamentally suppressing false trigger events. More importantly, the adjustment amplitude is not fixed, but is self-adaptive through the "jitter-certainty" joint mapping: under high certainty requirement tasks, the system has "zero tolerance" for jitter, and will use a larger gain coefficient to quickly raise the threshold to strictly ensure cooperative accuracy; in scenarios where some uncertainty is tolerated, a relatively moderate adjustment is adopted to avoid response delay of legitimate trigger signals due to excessively high threshold. This ensures that the handshake action between the intelligent analysis and compression encoding modules on board maintains a high degree of timing accuracy and cooperative certainty, even in complex space interference environments, laying a solid foundation for the reliable operation of the entire on-board processing link.

[0049] As Figure 4The diagram shows the structure of a low-Earth orbit satellite observation data compression and encoding processing system provided in this application embodiment. It includes: a task-aware analysis and collaborative compression module, a task performance-driven control module, a collaborative state monitoring and evaluation module, and a deterministic constraint-based collaborative control module. The task-aware analysis and collaborative compression module performs intelligent analysis of observation data in orbit, generating target detection results and region-of-interest masks containing semantic information. Based on the target detection results and region-of-interest masks, it performs task-aware differentiated compression and encoding on the original observation data to obtain task performance data. This data quantifies the efficiency of converting low-Earth orbit satellite observation data into effective information from the initial acquisition of original observation data from the satellite to the final generation of preset application requirements on the ground, resulting in an end-to-end task performance evaluation value. The task performance-driven control module determines whether to perform intelligent task performance control based on the end-to-end task performance evaluation value, achieving data fidelity... The shift from a task-driven to a mission-utility-driven approach achieves optimal synergy between information processing efficiency and mission success rate. If executed, the downlink bandwidth utilization and on-board processing pipeline latency are monitored in real time after adjustment; otherwise, the downlink bandwidth utilization and on-board processing pipeline latency are directly monitored in real time. The collaborative status monitoring and evaluation module acquires collaborative processing data during real-time monitoring of downlink bandwidth utilization and on-board processing pipeline latency. This data is used to quantify the collaborative accuracy of on-orbit intelligent analysis and compression coding collaboration on the low-Earth orbit satellite, obtaining on-board collaborative processing determinism. The deterministic constraint collaborative adjustment module determines whether to perform intelligent adjustment of collaborative processing accuracy based on the on-board collaborative processing determinism, thereby maintaining the reliability of the on-board processing link and the credibility of the output results. If executed, the compressed coding processing result is output after adjustment; otherwise, the compressed coding processing result is directly output.

[0050] In the embodiment, through the linkage of the task perception analysis and collaborative compression module, the task performance driven regulation module, the collaborative state monitoring and evaluation module, and the collaborative regulation module with deterministic constraints, the on-orbit data processing of the low-orbit satellite is fundamentally changed from "data fidelity driven" to "task utility driven", and the conversion efficiency of the on-orbit observation data to effective information on the ground and the reliability and stability of the on-orbit processing link are significantly improved. Among them, the task perception analysis and collaborative compression module generates target detection results and ROI masks through on-orbit intelligent analysis, and implements differential compression coding combined with task requirements, accurately mines the task value of observation data, effectively improves the end-to-end task performance evaluation value, and guarantees the efficient landing of ground preset application requirements (such as target identification report and disaster assessment map); the task performance driven regulation module dynamically decides whether to execute intelligent regulation based on the end-to-end task performance evaluation value, taking into account the collaborative optimization of information processing efficiency and task success rate, and avoiding the waste of bandwidth and the redundancy of computing power caused by excessive pursuit of data fidelity; the collaborative state monitoring and evaluation module accurately quantifies the collaborative accuracy of on-orbit intelligent analysis and compression coding based on real-time monitoring of the downlink bandwidth occupancy rate and the on-orbit processing pipeline delay state, providing a core basis for the stability evaluation of the on-orbit processing link; the collaborative regulation module with deterministic constraints dynamically executes collaborative accuracy regulation based on the on-orbit collaborative processing, effectively avoiding problems such as processing result distortion and unreliable link caused by collaborative deviation, and guaranteeing the reliability and consistency of the on-orbit processing output results. The four modules form a closed-loop link of perception-compression-evaluation-regulation, which not only maximizes the task value of the low-orbit satellite observation data, but also ensures the long-term stability and reliability of the on-orbit processing link.

[0051] The various features and processes described above can be used independently of one another or can be combined in various ways. All possible combinations and subcombinations are intended to fall within the scope of the present disclosure. In addition, some of the method or process blocks can be omitted in some implementations. The methods and processes described herein are also not limited to any particular order or sequence, and the blocks or states relating thereto can be performed in other suitable orders or sequences. For example, blocks or states described in succession can be performed at the same time or can be performed in an order that is different from the order described. Multiple blocks or states can be combined in a single block or state. The example blocks or states can be performed serially, in parallel, or in some other order. Blocks or states can be added to or removed from the disclosed example embodiments. The example systems and components described herein can be configured differently than described. For example, elements can be added to, removed from, or rearranged in the disclosed example embodiments.

[0052] The various operations of example methods described herein can be performed, at least partially, by an algorithm. The algorithm can be embodied in program code or instructions stored in a memory (e.g., the non-transitory computer-readable storage medium described above). Such an algorithm can include a machine learning algorithm. In some embodiments, a machine learning algorithm can not explicitly program a computer to perform a function, but can learn from training data to make a predictive model that performs the function.

[0053] The various operations of example methods described herein can be performed, at least partially, by one or more processors that are temporarily configured (e.g., by software) or permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors can constitute processor- implemented engines that operate to perform one or more operations or functions described herein.

[0054] Similarly, the methods described herein can be at least partially processor- implemented, with a particular processor or processors being an example of hardware. For example, at least some of the operations of a method can be performed by one or more processors or processor-implemented engines. Moreover, a processor or processors can also operate to support performance of the relevant operations in a "cloud computing" environment or as a "software as a service" (SaaS). For example, at least some of the operations can be performed by a group of computers (as examples of machines including processors), with these operations being accessible via a network (e.g., the Internet) and via one or more appropriate interfaces (e.g., an application program interface (API)).

[0055] The performance of certain of the operations can be distributed among the processors, not only residing within a single machine, but deployed across a number of machines. In some example embodiments, the processors or processor-implemented engines can be located in a single geographic location (e.g., within a home environment, an office environment, or a server farm). In other example embodiments, the processors or processor-implemented engines can be distributed across a number of geographic locations.

[0056] In this specification, a plurality of instances can implement a component, operation, or structure described as a single instance. Although individual operations of one or more methods are illustrated and described as separate operations, one or more of the individual operations can be performed concurrently, and nothing requires that the operations be performed in the order illustrated. Structures and functionality presented as separate components in example configurations can be implemented as a combined structure or component. Similarly, structures and functionality presented as a single component can be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of the subject matter herein.

[0057] While the overview of the subject matter has been described with reference to particular examples, various modifications and changes can be made to these embodiments without departing from the broader scope of the embodiments of the present disclosure. Such embodiments of the inventive subject matter can be referred to herein, individually or collectively, by the term "application" merely for convenience and without intending to voluntarily limit the scope of this application to any single disclosure or concept if indeed more than one disclosure or concept is, in fact, disclosed.

[0058] The embodiments illustrated herein are described in sufficient detail to enable those skilled in the art to practice the disclosed teachings. Other embodiments can be utilized and derived therefrom, such that structural and logical substitutions and changes can be made without departing from the scope of the present disclosure. The detailed description is, therefore, not to be taken in a limiting sense. The scope of the embodiments is defined by the appended claims, along with the full range of equivalents to which such claims are entitled.

Claims

1. A method for compressing and encoding low-orbit satellite observation data, characterized in that, Includes the following steps: S1. Perform intelligent analysis of observation data in orbit to generate target detection results and region of interest masks containing semantic information. Based on the target detection results and region of interest masks, perform task-aware differential compression encoding on the original observation data to obtain task performance data. This data is used to quantify the efficiency of converting low-orbit satellite observation data into effective information from the time the satellite acquires the original observation data to the time the preset application requirements are generated on the ground, and obtain an end-to-end task performance evaluation value. S2, based on the end-to-end task performance evaluation value, determine whether to perform intelligent task performance regulation, realizing the transformation from data fidelity-driven to task performance-driven, thereby achieving the optimal synergy between information processing efficiency and task success rate. If executed, the downlink bandwidth utilization and on-board processing pipeline latency status are monitored in real time after regulation. If not, the downlink bandwidth utilization and on-board processing pipeline latency status are directly monitored in real time. S3 acquires collaborative processing data during real-time monitoring of downlink bandwidth utilization and on-board processing pipeline latency. This data is used to quantify the collaborative accuracy of on-orbit intelligent analysis and compression coding collaboration on low-Earth orbit satellites, thereby obtaining on-board collaborative processing determinism. S4. Based on the deterministic determination of the on-board collaborative processing, whether to perform intelligent adjustment of the accuracy of collaborative processing is to maintain the reliability of the on-board processing link and the credibility of the output results. If it is performed, the compressed encoding processing result is output after adjustment; otherwise, the compressed encoding processing result is output directly.

2. The compression encoding processing method for low-orbit satellite observation data as described in claim 1, characterized in that, The specific steps for obtaining the end-to-end task performance evaluation value are as follows: The mission performance data includes the downlink transmission latency of low-Earth orbit satellite observation data, the end-to-end information utility bandwidth ratio, and the compression latency of low-Earth orbit satellite observation data. The transmission delay correction factor is combined with the ratio of the transmission delay threshold to the downlink transmission delay of low-Earth orbit satellite observation data to obtain the transmission delay impact component. The bandwidth ratio correction factor is combined with the ratio of the end-to-end information utility bandwidth ratio to the bandwidth ratio critical value to obtain the bandwidth ratio influence component. The compression delay correction factor is combined with the ratio of the compression delay to the compression delay threshold of the low-orbit satellite observation data to obtain the compression delay influence component. The transmission delay impact component, bandwidth ratio impact component, and compression delay impact component are coupled and processed to obtain the end-to-end task performance evaluation value. The specific steps for determining whether to perform intelligent task performance control are as follows: If the end-to-end task performance evaluation value is greater than or equal to the task performance benchmark value, then intelligent task performance control will not be performed; otherwise, dynamic window scheduling based on satellite-ground coordination and adaptive background compression based on channel congestion awareness will be performed to meet task timeliness.

3. The compression encoding processing method for low-orbit satellite observation data as described in claim 2, characterized in that, The specific steps for performing task-time-oriented space-ground collaborative dynamic window scheduling are as follows: For each low-Earth orbit satellite observation mission, a maximum end-to-end allowable delay threshold is predefined. The maximum end-to-end allowable delay threshold represents the maximum total time allowed from the time of low-Earth orbit satellite observation data acquisition to the time of final delivery to the end user. Before and during each satellite transit, the current ground processing and distribution latency is monitored and acquired in real time. Based on the end-to-end maximum allowable latency threshold and the ground processing and distribution latency, a total time window available for satellite-ground collaboration is obtained for on-board processing and downlink transmission. The total time window available for satellite-ground collaboration represents the difference between the end-to-end maximum allowable latency threshold and the ground processing and distribution latency.

4. The compression encoding processing method for low-orbit satellite observation data as described in claim 3, characterized in that, The execution of mission-time-oriented space-ground collaborative dynamic window scheduling also includes: Based on the end-to-end mission performance evaluation value and the total time window, combined with the preset time window for satellite transit, the adjusted actual scheduling and transmission time window is dynamically determined and instructed to be executed by the satellite through the following logic: If the total time window is greater than or equal to the preset time window, it indicates that the ground processing margin is sufficient, and the satellite can be instructed to carry out low-orbit satellite observation data transmission in the most stable manner within the entire preset transit time window. That is, the actual scheduled transmission time window is the preset time window. If the total time window is less than the preset time window, it indicates that the total time is tight. The transmission window is compressed to reserve enough time for ground processing. Based on the end-to-end mission performance evaluation value and the current total time window, it is input into the predefined distribution delay-transmission time window mapping relationship, outputs the actual scheduled transmission time window, and instructs the satellite to transmit high-priority data only within the actual scheduled transmission time window. The start and end times of the actual scheduled transmission time window, along with the corresponding data transmission priority strategy, are encapsulated into a scheduling instruction. When the satellite enters the previous ground station's visible window, it is instructed to execute low-orbit satellite observation data transmission according to the adjusted actual scheduled transmission time window.

5. The compression encoding processing method for low-orbit satellite observation data as described in claim 2, characterized in that, The specific steps of the adaptive background compression based on channel congestion awareness are as follows: The downlink total bandwidth utilization is monitored in real time. Based on the real-time monitored downlink total bandwidth utilization and the end-to-end task performance evaluation value, the maximum allowable compression intensity of the background area is dynamically adjusted. Specifically, if the downlink total bandwidth utilization is less than or equal to the bandwidth utilization threshold, it is determined that the channel idleness is high and the downlink total bandwidth resources are sufficient, and the maximum allowable compression intensity of the current background area is maintained. If the downlink total bandwidth utilization rate is greater than the bandwidth utilization rate threshold, the channel is determined to be congested. Based on the downlink total bandwidth utilization rate and the end-to-end task performance evaluation value, the input is fed into the predefined utilization rate-maximum compression intensity mapping relationship, and the maximum compression intensity of the target background area is output. The maximum compression is applied to the background area to reduce its data volume and reserve bandwidth for critical information.

6. The compression encoding processing method for low-orbit satellite observation data as described in claim 1, characterized in that, The specific steps to obtain the deterministic nature of on-board collaborative processing are as follows: The collaborative processing data includes the effective value of task performance, the latency of ROI content discrimination, and the collaborative completion rate of the over-the-top window. The effective influence component is obtained by combining the ratio of the effective value of task effectiveness to the effective threshold value of effectiveness with the effective correction factor. The discrimination delay correction factor is combined with the ratio of the discrimination delay threshold to the ROI content discrimination delay to obtain the discrimination delay impact component; The collaborative completion rate correction factor is combined with the ratio of the collaborative completion rate to the critical value of the collaborative completion rate in the top window to obtain the collaborative completion rate influence component. By coupling the effective impact component of performance, the impact component of discrimination delay, and the impact component of collaborative completion rate, the deterministic nature of on-board collaborative processing is obtained. The specific steps for determining whether to perform intelligent adjustment of the accuracy of collaborative processing are as follows: If the on-board collaborative processing determinism is greater than or equal to the collaborative processing baseline value, then intelligent adjustment of collaborative processing accuracy will not be performed; otherwise, adaptive queue adjustment based on processing delay awareness and adaptive trigger noise tolerance adjustment based on jitter awareness will be performed.

7. The compression encoding processing method for low-orbit satellite observation data as described in claim 6, characterized in that, The specific steps for performing adaptive queue control based on processing latency awareness are as follows: In the onboard intelligent processing pipeline, the ROI mask generation latency of the current frame is monitored in real time. Based on the real-time monitored ROI mask generation latency and the determinism of onboard collaborative processing, the overflow prevention waterline and data discarding strategy of the queue are dynamically adjusted, specifically as follows: If the ROI mask generation delay is less than or equal to the generation delay baseline value, the intelligent processing is determined to be running normally. The current on-board buffer queue management and overflow rate are set to the on-board buffer queue management and overflow rate threshold to allow the data stored in the queue to smooth downstream transmission fluctuations. If the ROI mask generation delay is greater than the generation delay baseline, intelligent processing will be delayed. Based on the determinism of ROI mask generation delay and on-board collaborative processing, the data is input into a predefined generation delay-overflow rate mapping relationship. The target on-board buffer queue management and overflow rate are output, and the on-board buffer queue management and overflow rate are adjusted downward to trigger data flow in advance, thus reserving buffer space for the backlog of low-orbit satellite observation data that may be caused by processing deceleration.

8. The compression encoding processing method for low-orbit satellite observation data as described in claim 6, characterized in that, The specific steps of the adaptive trigger noise tolerance control based on jitter perception are as follows: A predefined benchmark value for collaborative trigger delay jitter is defined. During on-board collaborative processing, the actual delay from analysis completion to compression trigger is collected and recorded each time. The current jitter statistics of the actual delay sequence within a preset window are calculated in real time. Based on the determinism of on-board collaborative processing and the current jitter statistics, the noise margin threshold used to determine the signal completion of analysis is dynamically adjusted. Specifically: If the current jitter statistic is less than or equal to the collaborative trigger delay jitter baseline value, the collaborative trigger timing is determined to be stable, and the current trigger threshold noise margin threshold is maintained.

9. The compression encoding processing method for low-orbit satellite observation data as described in claim 8, characterized in that, The adaptive trigger noise tolerance control based on jitter perception also includes: If the current jitter statistic is greater than the collaborative trigger delay jitter benchmark value, it is determined that there is abnormal jitter in the collaborative trigger timing. The trigger threshold noise margin threshold is increased to suppress false triggering events such as premature or delayed triggering caused by noise. Specifically, based on the on-board collaborative processing determinism and the current jitter statistic, it is input into the predefined delay jitter-margin threshold mapping relationship, and the output adjustment gain coefficient is used. The adjustment gain coefficient is combined with the trigger threshold noise margin threshold to obtain the target trigger threshold noise margin threshold, thereby raising the threshold for trigger determination and ensuring the accuracy and determinism of the collaborative action in timing.

10. A compression coding and processing system for low-Earth orbit satellite observation data, wherein the compression coding and processing system for low-Earth orbit satellite observation data is used to implement the compression coding and processing method for low-Earth orbit satellite observation data as described in any one of claims 1-9, characterized in that, The system includes: a task perception analysis and collaborative compression module, a task performance-driven control module, a collaborative state monitoring and evaluation module, and a collaborative control module with deterministic constraints. The task-aware analysis and collaborative compression module is used to perform intelligent analysis of observation data in orbit, generate target detection results and region of interest masks containing semantic information, and perform task-aware differentiated compression encoding on the original observation data based on the target detection results and region of interest masks to obtain task performance data. This data is used to quantify the efficiency of converting low-orbit satellite observation data into effective information from the time the satellite acquires the original observation data to the time the preset application requirements are generated on the ground, and to obtain an end-to-end task performance evaluation value. The task performance-driven control module is used to determine whether to perform intelligent task performance control based on the end-to-end task performance evaluation value. This realizes the transformation from data fidelity-driven to task performance-driven, thereby achieving the optimal synergy between information processing efficiency and task success rate. If the control is performed, the downlink bandwidth utilization and the latency status of the on-board processing pipeline are monitored in real time after the control is implemented. If not, the downlink bandwidth utilization and the latency status of the on-board processing pipeline are monitored directly in real time. The collaborative status monitoring and evaluation module is used to acquire collaborative processing data during real-time monitoring of downlink bandwidth utilization and on-board processing pipeline latency. This data is used to quantify the collaborative accuracy of on-orbit intelligent analysis and compression coding collaboration on low-Earth orbit satellites, and to obtain on-board collaborative processing determinism. The deterministic constraint-based collaborative control module is used to determine whether to perform intelligent control of the accuracy of the collaborative processing based on the deterministic judgment of the on-board collaborative processing, thereby maintaining the reliability of the on-board processing link and the credibility of the output results. If it is executed, the compressed encoding processing result is output after the control; otherwise, the compressed encoding processing result is output directly.

Images