Geographic information acquisition method and device based on remote sensing image, and medium

Abstract

The invention discloses a geographic information collection method and device based on a remote sensing image and a medium, and relates to the technical field of geographic information processing, and the method comprises the steps: an edge node obtains remote sensing image data, evaluates a self computing resource state, and generates a resource demand description; the edge node serves as a task publisher, splits the resource demand description into a plurality of sub-task contracts and broadcasts the sub-task contracts to the peripheral node through the equipment direct connection network; and after verifying the correctness and integrity of all subtask contract result data, the task publisher generates a geographic information product, uploads the geographic information product to the cloud for storage, pays the computing power reward to the corresponding resource provider, submits a transaction record to the cloud, and updates the historical reputation value of the corresponding resource provider. According to the method, accurate scheduling and efficient cooperation of remote sensing image processing tasks are realized through edge computing resource dynamic perception and an intelligent contract task decomposition mechanism.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of geographic information processing, and in particular to a geographic information collection method based on remote sensing images, equipment and a medium. BACKGROUND

[0002] Remote sensing image geographic information collection technology has gradually evolved from centralized cloud processing to distributed edge computing to meet the real-time processing needs of massive image data. The edge computing framework utilizes the resources of nodes close to the data source and combines device direct network to build a temporary communication topology, providing a basic framework for the rapid distribution and parallel processing of tasks. At the same time, the introduction of blockchain smart contract technology in distributed systems provides an automated execution and verification mechanism for resource transactions and trusted collaboration between edge nodes, promoting the development of decentralized computing models. The reputation evaluation system, as a key element in the distributed system, its dynamic updating mechanism is also applied to encourage the reliable participation of nodes and constrain malicious behavior.

[0003] However, the existing technical solutions still have significant bottlenecks in the distributed task scheduling and resource matching link in the edge computing environment. In particular, in a heterogeneous and highly dynamic resource changing edge node network, there is a lack of an efficient quantitative decision model that can consider the multi-dimensional real-time state of nodes and the multi-dimensional attributes of tasks. This makes it difficult for task publishers to quickly filter out the most optimal collaboration nodes in complex and changing bidding responses, which can easily lead to task allocation bias, response delay or resource mismatch, ultimately affecting the overall generation efficiency of geographic information products and system robustness. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a geographic information collection method based on remote sensing images, equipment and a medium, which solves the problem of low efficiency of distributed task matching caused by dynamic resource state and reputation maintenance demand of heterogeneous nodes in the edge computing environment.

[0006] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a geographic information acquisition method based on remote sensing images, which comprises: an edge node acquires remote sensing image data and evaluates its own computing resource state, and generates a resource demand description; the edge node as a task publisher splits the resource demand description into a plurality of sub-task contracts and broadcasts them to surrounding nodes through a device direct connection network; each sub-task contract contains a computing power reward; a surrounding node receiving the broadcast as a resource provider makes a multi-attribute decision game according to the local resource state and the historical credit value, and returns a bid response containing the bid, the historical credit value and the capability proof to the task publisher; the task publisher executes smart contract matching based on the bid and the historical credit value in the bid response, selects the resource provider and establishes a point-to-point computing collaboration channel; the resource provider acquires the remote sensing image data to be processed through the point-to-point computing collaboration channel and performs distributed computing, and returns the sub-task contract result data and the execution proof to the task publisher; after the task publisher verifies the correctness and integrity of all the sub-task contract result data, the geographic information product is generated and uploaded to the cloud storage, and at the same time, the computing power reward is paid to the corresponding resource provider and the transaction record is submitted to the cloud, and the historical credit value of the corresponding resource provider is updated.

[0007] As a preferred scheme of the geographic information acquisition method based on remote sensing images, the resource demand description includes the data size of the remote sensing image data to be processed, the required computing accuracy level, the maximum allowed delay of the task, and the maximum computing power reward total amount that the task publisher is willing to pay.

[0008] As a preferred scheme of the geographic information acquisition method based on remote sensing images, the resource demand description is split into a plurality of sub-task contracts and broadcasted to surrounding nodes through a device direct connection network, and the specific steps are as follows: The resource demand description is divided into a plurality of independent processing blocks, each independent processing block is assigned a computing complexity weight and a corresponding computing power reward allocation proportion; The sub-task contract is generated according to the computing complexity weight and the computing power reward allocation proportion; The task publisher broadcasts the sub-task contract to the surrounding nodes within the communication range through the device direct connection network.

[0009] As a preferred scheme of the geographic information acquisition method based on remote sensing images, the multi-attribute decision game refers to that the resource provider constructs a multi-dimensional decision vector based on the computing power reward attractiveness value, the task urgency value, the remaining power state and the historical credit value maintenance demand, calculates the comprehensive bid value through weighted scoring, and generates the optimal bid strategy.

[0010] As a preferred scheme of the remote sensing image-based geographic information collection method, the smart contract matching refers to that the task publisher establishes a two-dimensional evaluation matrix based on the bid and the historical credit value, and selects the resource provider from all the bid responses by using a Pareto optimal screening mechanism.

[0011] As a preferred scheme of the remote sensing image-based geographic information collection method, the selected resource provider establishes a point-to-point computing collaboration channel, and the specific steps are as follows: The task publisher sends a cooperation confirmation instruction to the selected resource provider, and the resource provider returns an encrypted identity authentication credential. The task publisher generates a temporary encryption key after verifying the identity authentication credential. An end-to-end encrypted point-to-point computing collaboration channel is established based on the temporary encryption key.

[0012] As a preferred scheme of the remote sensing image-based geographic information collection method, the distributed computing execution refers to that the resource provider performs feature extraction and semantic segmentation on the remote sensing image data according to the sub-task contract, and generates an execution proof containing a timestamp and a digital signature.

[0013] As a preferred scheme of the remote sensing image-based geographic information collection method, the historical credit value of the corresponding resource provider is updated, and the specific steps are as follows: The cloud receives the transaction record submitted by the task publisher and analyzes the validity of the execution proof. The credit gain value of this transaction is calculated according to the timeliness of the sub-task contract completion. The existing historical credit value of the resource provider is retrieved, and the historical credit value and the credit gain value are weighted and accumulated to update the historical credit value.

[0014] In a second aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and wherein the computer program is executed by the processor to implement any step of the remote sensing image-based geographic information collection method according to the first aspect of the present application.

[0015] In a third aspect, the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement any step of the remote sensing image-based geographic information collection method according to the first aspect of the present application.

[0016] The present application has the beneficial effects that: through the edge computing resource dynamic perception and intelligent contract task decomposition mechanism, the accurate scheduling and efficient cooperation of remote sensing image processing tasks are realized; a multi-dimensional node evaluation strategy fusing economy and reliability is adopted to construct a high-trust geographic information collection network; and relying on the double protection of encrypted communication and dynamic reputation system, the data security and processing quality are ensured, and a self-optimized distributed geographic information collection ecology is formed. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used 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.

[0018] Figure 1 Flowchart of the geographic information collection method based on remote sensing image.

[0019] Figure 2 Flowchart of subtask contract generation and broadcast process.

[0020] Figure 3 Flowchart of multi-attribute decision game process.

[0021] Figure 4 Flowchart of intelligent contract matching and collaboration channel establishment. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification.

[0023] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0024] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0025] Reference Figure 1 , for one embodiment of the present application, the embodiment provides a geographic information collection method based on remote sensing image, comprising the following steps: S1: The edge node obtains remote sensing image data and evaluates its own computing resource status to generate a resource requirement description; the resource requirement description includes the data volume of the remote sensing image data to be processed, the required computing accuracy level, the maximum allowable task delay, and the maximum total computing power reward that the task publisher is willing to pay.

[0026] For details, please refer to Figure 2 , the edge node reads the remote sensing image data to be processed from the connected remote sensing image sensor or local storage device; the specific operations include accessing the sensor data stream or local file and extracting the remote sensing image data file.

[0027] The edge node calls the built-in resource monitoring interface to query the current computing resource status, including available CPU computing power, free random access memory capacity, available storage space and remaining battery power; the query operation is performed through the standard application programming interface.

[0028] Based on a predefined task profile, the edge node loads the required computational accuracy level, the maximum allowable task latency, and the maximum total computing power reward the task issuer is willing to pay. This predefined task profile is stored locally on the edge node and contains fixed parameter values. A predefined task profile is a configuration file stored locally on the edge node that contains fixed parameter values ​​for the required computational accuracy level, the maximum allowable task latency, and the maximum total computing power reward the task issuer is willing to pay. The edge node generates a resource requirement description based on the data volume of the remote sensing imagery, the state of the computing resources, and the parameter values.

[0029] S2: The edge node acts as a task publisher, splitting the resource requirement description into several subtask contracts and broadcasting them to surrounding nodes through the device direct connection network.

[0030] S2.1: Divide the resource requirement description into several independent processing blocks, assign a computational complexity weight to each independent processing block, and determine the corresponding computing power reward distribution ratio.

[0031] Specifically, the task publisher calls the standard image processing function library (such as the image segmentation function in OpenCV) to divide the remote sensing image data into fixed-size image rectangular blocks; each image rectangular block is an independent processing block, ensuring that the data is spatially continuous and can be processed in parallel.

[0032] It should be noted that each independent processing block (such as the k-th image rectangle) will generate a unique corresponding subtask contract; the subtask contract embeds the computational complexity weight and reward distribution ratio of the independent processing block; k is the index variable of the image rectangle.

[0033] Furthermore, the task publisher assigns a computational complexity weight to each independent processing block.

[0034] Specifically, the complexity weight is calculated based on the data size of the independent processing block and the required computational accuracy level in the resource requirement description.

[0035] Calculate the complexity weight, the expression is: ; Where, Indicates the The computational complexity weight of each independent processing block, Indicates the The data size of each independent processing block, The required calculation accuracy level The scaling factor, An index variable representing an independent processing block.

[0036] It should be noted that The required calculation accuracy level The scaling factor is used to quantify the impact of different computational precision levels on the complexity of the processing task, specifically based on the precision level The weight values ​​are dynamically adjusted to reflect the additional computational resources required for high-precision tasks.

[0037] For example, higher precision algorithms (such as using double-precision floating point numbers instead of single precision) will increase the number of computational iterations or memory usage, thereby increasing the overall complexity. Usually preset by the task configuration file to discrete values ​​(such as Indicates low precision, Indicates medium precision, Indicates high precision, 4 for ultra-high precision), scaling factor Then through linear or nonlinear mapping and Association, for example, in a common implementation, when from Increase to hour, The value range can be set to to , for example low precision ( )correspond (baseline complexity), medium precision ( )correspond (complexity increased by 50%), high precision ( )correspond =2.0 (complexity doubled), while ultra-high precision ( 4) Can be expanded to This allows for sub-pixel analysis of more complex scenarios, such as remote sensing images, ensuring that weight calculations accurately match actual resource requirements, avoiding insufficient or wasted resource allocation.

[0038] The task publisher determines the computing power reward distribution ratio for each independent processing block; the computing power reward distribution ratio is based on the computational complexity weight.

[0039] The expression for computing power reward distribution ratio is: ; Where, For the The distribution ratio of computing power rewards for independent processing blocks indicates the weight of the task in the total computing tasks. represents the sum of the computational complexity weights of all independent processing blocks, Indicates the total number of independent processing blocks, Indicates summing over variables.

[0040] S2.2: Generate subtask contracts based on the computational complexity weight and computing power reward distribution ratio.

[0041] Specifically, the task publisher generates subtask contracts based on the complexity weight and computing power reward distribution ratio. Each subtask contract contains computing power reward and completion time limit, task identifier, data block identifier, calculation accuracy requirements, data space range description and verification rules.

[0042] The task publisher generates a unique task identifier for each independently processed block; the task identifier is composed of the publisher node identifier, timestamp and block index to ensure global uniqueness; According to the computational complexity weight and the computing power reward distribution ratio, the computing power reward of each subtask contract is calculated. The expression is: ; Where, For the The computing power reward of each subtask contract represents the reward that the resource provider can obtain after completing the corresponding task. For the The computing power reward distribution ratio of each subtask contract, Indicates the maximum total amount of computing power rewards that the task publisher is willing to pay in the resource requirement description. The index variable representing the subtask contract; At the same time, according to the maximum allowable delay of the task in the resource requirement description and the computational complexity weight of the independent processing block , assign the completion time limit of each subtask contract, the expression is: ; wherein, is the completion deadline of the thsub-task contract, represents the start timestamp of the task; The task publisher encapsulates the computing power reward, completion deadline, task identifier, data block identifier, computing accuracy requirement, data space range description, and verification rule into structured data to generate a sub-task contract.

[0043] The verification rule includes data integrity check code (such as SHA-256 hash value) and result data format specification, which is used for subsequent verification of correctness.

[0044] S2.3: The task publisher broadcasts the sub-task contract to the surrounding nodes within the communication range through the device direct connection network.

[0045] Specifically, the task publisher broadcasts the sub-task contract to the surrounding nodes within the communication range through the device direct connection network (such as Wi-Fi Direct protocol). The specific operation is: the task publisher uses the wireless communication interface to encapsulate the sub-task contract into a broadcast data packet, and sends it to the neighboring nodes through the broadcast address, ensuring that the broadcast data packet contains the complete content of the sub-task contract (such as computing power reward, completion deadline, and related metadata).

[0046] Preferably, the resource demand is split into sub-task contracts and broadcasted through the device direct connection network, realizing decentralized and fast matching, avoiding the risk of single point failure.

[0047] S3: The surrounding nodes receiving the broadcast serve as resource providers, and conduct multi-attribute decision game according to the local resource state and historical reputation value, and return a bidding response containing the bid, historical reputation value, and capability proof to the task publisher.

[0048] It should be noted that, referring to Figure 3 , multi-attribute decision game refers to that the resource provider constructs a multi-dimensional decision vector based on the computing power reward attractiveness value, task urgency value, self remaining power state, and historical reputation value maintenance demand, calculates the comprehensive bidding value through weighted scoring, and generates the optimal bidding strategy.

[0049] S3.1: The resource provider extracts the computing power reward, completion deadline, and related metadata of the sub-task contract from the broadcast data packet; calls the resource monitoring interface to obtain the current available central processing unit computing power, idle random access memory capacity, and remaining battery power, and accesses the cloud storage through the encrypted network connection, uses the device unique identifier to query and download the current historical reputation value.

[0050] Related metadata refers to the supplementary parameters required by resource providers to perform computing tasks in addition to computing power rewards and completion deadlines in the subtask contract. Specifically, they include: independent processing block identification, computing accuracy requirements, and data space range description (geographic boundary information after image segmentation).

[0051] S3.2: Define a four-dimensional decision vector; the four-dimensional decision vector includes the attractiveness of computing power rewards, the remaining power status of the task urgency, and the historical reputation value maintenance requirements.

[0052] Computing power-based rewards and local computing power costs The ratio of , calculates the attractiveness of computing power reward , the expression is: ; Among them, the local computing power cost refers to the comprehensive converted cost of hardware resources and energy expenses consumed by the device for computing and processing when the resource provider executes the subtask contract; hardware resources include CPU and memory, etc.; energy expenses include battery loss, etc.

[0053] Calculate the task urgency value, the expression is: ; Where, Indicates the urgency value of the task. The completion time limit specified in the subtask contract, The current timestamp of the resource provider. is a very small constant (such as 0.001) used to prevent the denominator from being zero.

[0054] The remaining battery status is obtained directly from the resource monitoring interface in the form of a remaining battery percentage.

[0055] Based on the difference between the historical reputation value and the target reputation value, the historical reputation value maintenance requirement is calculated. The expression is: ; Where, Indicates the minimum reputation gain required for the resource provider to reach the target reputation value, represents the target reputation value, Represents the historical reputation value, is a non-negative truncation function.

[0056] It should be noted that the non-negative truncation function is a standard function in mathematics. Its function is to limit the calculation result to a non-negative value, ensuring that the credit maintenance demand is reset to zero when the historical credit value exceeds the target value.

[0057] Based on the four-dimensional decision vector, a weighted scoring formula is used to calculate the comprehensive bid value.

[0058] Specifically, the resource provider calculates a comprehensive bidding value reflecting its willingness and ability to participate by weighting the computing power reward attractiveness value, the task urgency value, the remaining power state and the historical reputation value maintenance demand four dimensions of evaluation indicators according to pre-set weights; the pre-set weights refer to the fixed weight proportion of each dimension indicator uniformly configured by the cloud in the initialization stage according to the device type, task historical performance and current running strategy; the fixed weight proportion remains unchanged within the task execution period.

[0059] If the comprehensive bidding value is greater than the pre-set bidding threshold, a bidding response is generated; if the comprehensive bidding value is less than or equal to the pre-set bidding threshold, the bidding is abandoned. The pre-set bidding threshold is a decision boundary value pre-set by the resource provider according to the device performance characteristics and running strategy; the decision boundary value is generated by analyzing historical task execution data and resource consumption rules, and is used to determine whether the current comprehensive bidding value meets the minimum requirement for participating in the task.

[0060] The bidding response or the abandonment of bidding is taken as a bidding response instruction package, and the bidding response instruction package is sent to the task publisher through the device direct connection network.

[0061] Preferably, the resource provider generates a bidding response based on multi-attribute decision game, and quantifies the computing power reward attractiveness value, the task urgency value, the remaining power and the reputation maintenance demand, which is more accurate in matching the actual willingness of the node than the traditional single bidding strategy.

[0062] S4: Please refer to Figure 4 , the task publisher performs smart contract matching based on the bid in the bidding response and the historical reputation value, selects the resource provider and establishes a point-to-point computing collaboration channel; the smart contract matching refers to that the task publisher establishes a two-dimensional evaluation matrix based on the bid and the historical reputation value, and selects the resource provider from all bidding responses by using a Pareto optimal screening mechanism; the Pareto optimal screening mechanism is realized by using a non-dominated sorting genetic algorithm.

[0063] S4.1: The task publisher sends a cooperation confirmation instruction to the selected resource provider, and the resource provider returns an encrypted identity verification credential.

[0064] The task publisher generates a two-dimensional evaluation vector for each bidding response; the two-dimensional evaluation vector includes the bid and the historical reputation value; the bid represents the bid value in the bidding response of the resource provider, and the smaller the value is, the lower the cost is; the historical reputation value represents the historical reputation value in the bidding response of the resource provider, and the larger the value is, the higher the reputation is.

[0065] The evaluation vectors of all bidding responses are taken as initial population bidding response individuals for non-dominated sorting.

[0066] Specifically, for any two bid response individuals and bid response individual , if both of the following conditions are met, bid response individual is said to dominate bid response individual ; bid response individual and bid response individual represent two independent bid schemes submitted by two different resource providers respectively, and there are distinguishable difference characteristics in the supplier identity, bid parameters or performance conditions.

[0067] It should be noted that the bid response individual comparison rule is as follows: the bid of bid response individual should be less than the bid of bid response individual ; the historical credit value of bid response individual should be greater than the historical credit value of bid response individual .

[0068] Further, through multiple rounds of non-dominated comparison, the bid response individuals are divided into a hierarchical sequence with clear superior and inferior relationships.

[0069] Specifically, the first layer (non-dominated solution set): all bid response individuals that are not dominated by any other bid response individual form the first layer, i.e. the optimal solution set; the second layer: all bid response individuals dominated by the bid response individuals of the first layer form the second layer; subsequent layers: in this way, until all bid response individuals are classified into a certain layer.

[0070] Through layer-by-layer screening, it is ensured that the bid response individuals of each layer are superior to the next layer, forming a Pareto frontier. Based on this Pareto frontier, the task publisher will preferentially select resource providers from the first layer (optimal solution set). However, the first layer (i.e. the Pareto frontier) itself usually contains multiple "non-dominated" solutions; "non-dominated" solutions have their own advantages in bid and credit and cannot be directly compared.

[0071] Further, in order to further screen in this optimal solution set, the crowding distance of the bid response individuals in the same non-dominated layer needs to be calculated, the purpose being to preferentially select individuals located in the sparse area of the solution set (i.e. more unique features of the scheme) to maintain the diversity of the selection.

[0072] It should be noted that when calculating the crowding distance of the bid response individuals in the same non-dominated layer, a k-nearest neighbor-based density estimation algorithm can be used, and the specific method is as follows: for each solution in the Pareto frontier, the Euclidean distance between it and the adjacent solution in the target space (the two-dimensional space formed by the bid and the historical credit value) is calculated, and the average of the distances of the adjacent two points in each dimension is taken as the crowding distance.

[0073] By combining the two indicators of non-dominated hierarchy and crowding distance, the bidding response individuals with high non-dominated hierarchy and largest crowding distance are selected to form the final optimal resource provider set.

[0074] The highest-ranked resource provider (such as the one with the largest congestion distance in the first layer) is selected from the optimal resource provider set, and the task publisher sends a cooperation confirmation instruction to the selected resource provider through the device direct connection network; in response to the cooperation confirmation instruction, the resource provider generates a digital certificate containing a unique device identifier and a random number, encrypts it using an asymmetric encryption algorithm (such as RSA), and returns it to the task publisher to complete the security authentication process.

[0075] The better approach is to use the Pareto optimal screening mechanism combined with the non-dominated sorting genetic algorithm for smart contract matching, giving priority to high-reputation nodes while ensuring low costs. Compared with the traditional bidding mechanism, it optimizes both economy and reliability.

[0076] S4.2: The task issuer generates a temporary encryption key after verifying the authentication credentials.

[0077] Specifically, the task publisher uses the resource provider's public key to decrypt the authentication credential and verify whether the device unique identifier is consistent with the bid response.

[0078] The task publisher calls a key generation function (such as the ECDH algorithm) to generate a temporary encryption key.

[0079] S4.3: Establish an end-to-end encrypted peer-to-peer computing collaboration channel based on a temporary encryption key.

[0080] Specifically, the task publisher and the resource provider establish an end-to-end encrypted peer-to-peer computing collaboration channel based on a temporary encryption key through a standard communication protocol (such as DTLS).

[0081] Based on a peer-to-peer computing collaborative channel, test messages are sent bidirectionally and the consistency of the decrypted content is verified to confirm that the channel is secure and available.

[0082] Better yet, it establishes an end-to-end encrypted peer-to-peer computing collaboration channel, which significantly reduces encryption communication overhead compared to traditional VPN tunnels and ensures data privacy during distributed computing processes.

[0083] S5: The resource provider obtains the remote sensing image data to be processed through the peer-to-peer computing collaboration channel and performs distributed computing, outputting the subtask contract result data and execution proof back to the task publisher; executing distributed computing means that the resource provider performs feature extraction and semantic segmentation on the remote sensing image data according to the subtask contract, and generates an execution proof containing a timestamp and digital signature.

[0084] S5.1: The resource provider receives the remote sensing image data to be processed sent by the task publisher through an end-to-end encrypted point-to-point computing collaboration channel (the processing of remote sensing image data corresponds to the independent processing block specified in the subtask contract).

[0085] The resource provider calls a preloaded image processing function library (such as OpenCV) to perform feature extraction, denoising and normalization on the remote sensing image data to be processed; uses a scale-invariant feature transformation algorithm to extract key point features in the remote sensing image; and generates a feature description vector set based on the key point features.

[0086] It should be noted that key point features refer to local structures in an image that are significantly distinguishable (such as corner points and edge intersections), which specifically include position, scale and direction information and are used to uniquely identify specific areas in the image.

[0087] S5.2: The resource provider uses a pre-trained lightweight convolutional neural network model (such as MobileNetV2) to perform semantic segmentation operations on remote sensing image data: classify each pixel into a predefined land feature category (such as water, vegetation, and buildings) and output a semantic segmentation mask map with geographic coordinates.

[0088] Combine feature description vector sets, semantic segmentation masks with geographic coordinates, and metadata (including spatial extent identifiers of independently processed blocks) into structured data.

[0089] Based on the structured data, the resource provider generates an execution certificate containing timestamp and digital signature elements; the timestamp calls the driver interface of the processor timestamp counter to obtain the precise time when the calculation is completed; the digital signature uses the resource provider's private key to encrypt and sign the hash value of the subtask contract result data.

[0090] Through the peer-to-peer computing collaboration channel, the subtask contract result data and execution proof are encapsulated into a data packet and sent to the task publisher.

[0091] S6: After the task publisher verifies the correctness and completeness of all subtask contract result data, it generates geographic information products and uploads them to cloud storage. At the same time, it pays computing power rewards to the corresponding resource providers and submits transaction records to the cloud, updating the historical reputation value of the corresponding resource providers.

[0092] S6.1: The cloud receives the transaction record submitted by the task publisher and analyzes the validity of the execution proof.

[0093] Specifically, the task publisher uploads all subtask contract result data and the corresponding execution proof, resource provider identity identifier, details of the computing power remuneration to be paid, and task global metadata (such as task identifier and geographical scope) to cloud storage.

[0094] Further, the cloud performs a verification operation: checks whether the timestamp in the execution proof is within the completion time limit specified by the subtask contract; decrypts the digital signature using the public key of the resource provider to obtain the hash value ; calculates the hash value of the subtask contract result data .

[0095] If , the verification is passed, and the reputation gain value of this transaction is calculated according to the timeliness of the completion of the subtask contract.

[0096] Specifically, the timeliness of the completion of the subtask contract is obtained by comparing the actual completion timestamp in the execution proof with the completion time limit specified by the subtask contract, and the specific calculation process is: obtaining the timestamp in the execution proof, and extracting the task start timestamp and the maximum allowed completion time limit specified in the subtask contract, calculating the actual time consumption; the proportion of the actual time consumption to the completion time limit is taken as the timeliness quantification basis, and the timeliness score is mapped according to the preset segmentation rule .

[0097] It should be noted that the preset segmentation rule divides three evaluation levels according to the proportion of the actual task completion time to the specified total time limit: when the resource provider actually consumes time earlier than half of the total time limit, the highest timeliness evaluation level is given; when the actual time consumption does not exceed the total time limit but exceeds half of it, the medium timeliness evaluation level is given; when the actual time consumption exceeds the total time limit, the timeliness evaluation level is zero. The segmentation rule quantifies the contribution degree of the speed to the timeliness through the ladder, ensuring that the enthusiasm of the resource provider to complete the task in advance is fully recognized, and the critical completion and overtime behaviors are gradiently distinguished and evaluated.

[0098] When the timeliness score , full score is obtained, indicating early completion, when , half score is obtained, indicating on-time completion, and when , zero score is obtained, indicating overtime.

[0099] According to the timeliness score , the reputation gain value of this transaction is calculated , and the expression is: ; In the formula, is a preset gain coefficient (example value is 0.1).

[0100] It should be noted that the preset gain coefficient is a fixed parameter preset by the cloud manager, which is used to quantify the influence strength of the quality of a single task completion on the reputation value of the resource provider; the cloud manager refers to a management entity responsible for maintaining the cloud storage, processing data interaction between the task publisher and the resource provider, and updating the reputation value (essentially a general service management subject in cloud computing).

[0101] The gain coefficient value is configured according to the update sensitivity requirement of the reputation system in the actual business scenario. For example, a higher value can be set to strengthen the incentive effect in a scenario that requires rapid reflection of node reliability, and a lower value can be set to smooth fluctuations in a scenario that emphasizes reputation stability.

[0102] The gain coefficient determines the maximum increase amplitude of the reputation value for a single task; for example, when the resource provider completes the task in advance , the reputation gain value of 0.1 can be obtained for this task, and if the task is completed on time , the reputation gain value of 0.1 can be obtained for this task, and if the task is completed on time , the reputation gain value of 0.1 can be obtained for this task, and if the task is completed on time , the reputation gain value of 0.1 can be obtained for this task, and if the task is completed on time

[0103] S6.2: Retrieve the existing historical reputation value of the resource provider, and perform weighted accumulation calculation on the historical reputation value and the reputation gain value to update the historical reputation value.

[0104] Specifically, the cloud queries the current historical reputation value stored in the database according to the resource provider identifier ; the database refers to a structured data set in the cloud specially used for recording and maintaining the historical reputation value of the resource provider; the database is based on a distributed key-value storage architecture (such as Redis or Cassandra), and the resource provider identifier is used as the unique key to realize efficient query and update operation, and ensure the reliability and real-time performance of the reputation value data.

[0105] The current historical reputation value The update formula is: ; In the formula, represents the updated historical reputation value, is the historical weight coefficient, which represents the weight of the original historical reputation value in the update calculation, is the reputation gain value of this transaction, which reflects the timeliness performance of the resource provider in completing the sub-task contract, represents the weight coefficient of the reputation gain value of this transaction.

[0106] It should be noted that the historical weight coefficient is preset as a fixed constant by the cloud manager, which is used to balance the weight proportion of the historical reputation value and the current contribution, and the example value range is to As ).

[0107] The embodiment also provides a computer device, comprising a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the remote sensing image based geographic information acquisition method provided in the above embodiment.

[0108] The computer device can be a terminal, and the computer device comprises a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved through WIFI, an operator network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0109] The embodiment also provides a storage medium having a computer program stored thereon, the program being executed by a processor to realize the remote sensing image based geographic information acquisition method provided in the above embodiment. The storage medium can be realized by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk.

[0110] To sum up, the application realizes accurate scheduling and efficient cooperation of remote sensing image processing tasks through edge computing resource dynamic perception and intelligent contract task decomposition mechanism; adopts a multi-dimensional node evaluation strategy that fuses economy and reliability to build a high-trust geographic information collection network; relies on dual protection of encrypted communication and dynamic reputation system to ensure data security and processing quality, forming a self-optimizing distributed geographic information collection ecology.

[0111] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and they should be covered in the scope of the claims of the present application.

Claims

1. A method for collecting geographic information based on remote sensing images, characterized by: include: Edge nodes acquire remote sensing image data and evaluate their own computing resource status to generate resource demand descriptions; As the task publisher, the edge node splits the resource requirement description into several subtask contracts and broadcasts them to surrounding nodes through the device direct connection network; each subtask contract contains computing power rewards; The surrounding nodes that receive the broadcast act as resource providers and conduct a multi-attribute decision-making game based on the local resource status and historical reputation value, and return a bidding response containing the quotation, historical reputation value and capability proof to the task publisher; The task publisher executes smart contract matching based on the bid response and historical reputation value, selects the resource provider and establishes a peer-to-peer computing collaboration channel; The resource provider obtains the remote sensing image data to be processed through the peer-to-peer computing collaboration channel and performs distributed computing, outputting the subtask contract result data and execution proof back to the task publisher; After the task publisher verifies the correctness and completeness of all subtask contract result data, it generates geographic information products and uploads them to cloud storage. At the same time, it pays computing power rewards to the corresponding resource providers and submits transaction records to the cloud, updating the historical reputation value of the corresponding resource providers.

2. The method for collecting geographic information based on remote sensing images according to claim 1, wherein: The resource requirement description includes the amount of remote sensing image data to be processed, the required level of computing accuracy, the maximum allowable delay of the task, and the maximum total amount of computing power compensation that the task publisher is willing to pay.

3. The method for collecting geographic information based on remote sensing images according to claim 1, wherein: The resource requirement description is split into several subtask contracts and broadcasted to surrounding nodes via the device direct connection network. The specific steps are as follows: Divide the resource demand description into several independent processing blocks, assign a computational complexity weight to each independent processing block and determine the corresponding computing power reward distribution ratio; Generate subtask contracts based on computational complexity weights and computing power reward allocation ratios; The task publisher broadcasts the subtask contract to surrounding nodes within the communication range through the device direct connection network.

4. The method for collecting geographic information based on remote sensing images according to claim 1, wherein: The multi-attribute decision-making game refers to the resource provider constructing a multi-dimensional decision vector based on the computing power reward attractiveness value, task urgency value, its own remaining power status and historical reputation value maintenance requirements, calculating the comprehensive bidding value through weighted scoring, and generating the optimal bidding strategy.

5. The method for collecting geographic information based on remote sensing images according to claim 1, wherein: The smart contract matching refers to the task publisher establishing a two-dimensional evaluation matrix based on quotations and historical reputation values, and using a Pareto optimal screening mechanism to select resource providers from all bidding responses.

6. The method for collecting geographic information based on remote sensing images according to claim 1, wherein: The specific steps of selecting a resource provider and establishing a peer-to-peer computing collaboration channel are as follows: The task publisher sends a cooperation confirmation instruction to the selected resource provider, and the resource provider returns an encrypted authentication credential; The task publisher verifies the authentication credentials and generates a temporary encryption key; Establish an end-to-end encrypted peer-to-peer computing collaboration channel based on temporary encryption keys.

7. The method for collecting geographic information based on remote sensing images according to claim 1, wherein: The execution of distributed computing refers to the resource provider performing feature extraction and semantic segmentation on remote sensing image data according to the subtask contract, and generating an execution certificate including a timestamp and digital signature.

8. The method for collecting geographic information based on remote sensing images according to claim 1, wherein: The specific steps for updating the historical reputation value of the corresponding resource provider are as follows: The cloud receives the transaction records submitted by the task publisher and analyzes the validity of the execution proof; Calculate the reputation gain value of this transaction based on the timeliness of the subtask contract completion; Retrieve the existing historical reputation value of the resource provider, perform weighted cumulative calculation on the historical reputation value and the reputation gain value, and update the historical reputation value.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for collecting geographic information based on remote sensing images according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for collecting geographic information based on remote sensing images according to any one of claims 1 to 8 are implemented.

Images