WO2023050745A1

WO2023050745A1 - Image processing method and apparatus, device, medium, and program

Info

Publication number: WO2023050745A1
Application number: PCT/CN2022/083772
Authority: WO
Inventors: 罗均; 徐翔; 许鑫汉
Original assignee: 上海商汤智能科技有限公司
Priority date: 2021-09-29
Filing date: 2022-03-29
Publication date: 2023-04-06
Also published as: CN113887396A

Abstract

Embodiments of the present disclosure provide an image processing method and apparatus, a device, a medium, and a program. The method is performed by an electronic device. The method comprises: acquiring an image to be processed; determining a plurality of computing node information and execution flow information on the basis of a target task corresponding to said image, wherein computing nodes respectively corresponding to the plurality of computing node information are used for completing different preset functions; generating, on the basis of the plurality of computing node information and the execution flow information, an execution flow corresponding to the execution flow information, wherein the execution flow comprises at least one computing node information; and generating calculation graph information on the basis of the execution flow, and processing said image according to the calculation graph information to obtain an image processing result. Thus, repeated storage of excessive redundant codes is reduced, the platform storage load is reduced, the use flexibility of the computing nodes is improved, and the management difficulty of a postprocessing algorithm is reduced.

Description

An image processing method, device, equipment, medium and program

Cross References to Related Applications

This patent application claims the priority of the Chinese patent application number 202111152501.8 submitted on September 29, 2021, the applicant is Shanghai Shangtang Intelligent Technology Co., Ltd., and the application name is "an image processing method, device, equipment, medium and program" , which is incorporated by reference in its entirety into this application.

technical field

The present disclosure relates to the technical field of image processing, and in particular to an image processing method, device, equipment, medium and program.

Background technique

There are many algorithm models for remote sensing image interpretation based on artificial intelligence (AI) (ie, AI remote sensing detection models), such as target detection, image segmentation, change detection, etc., and each AI remote sensing detection model corresponds to at least one set The post-processing algorithm will generate more customized post-processing functions when responding to different industry application requirements. For example, the post-processing of the building extraction model can include removing small image spots, regularizing the building extraction results, vector The post-processing of the water body recognition model can include functions such as removing small patches, blurring boundaries, growing boundaries, and fusion of multi-model results. According to the task requirements of different industry applications, it is often necessary to make appropriate adjustments to the post-processing algorithm, such as adding functions such as vectorization or removing small plaques in the post-processing algorithm.

In related technologies, when post-processing codes are often modified, many versions of post-processing algorithms for different tasks are formed. Different versions of post-processing algorithms may include codes with repeated functions, resulting in support for AI remote sensing detection models and post-processing algorithms. The platform where the processing algorithm runs stores excessive redundant codes, resulting in a large storage load on the platform. At the same time, it is difficult to uniformly manage different versions of the post-processing algorithm. In addition, every time the user needs to adjust the post-processing algorithm, R&D personnel are required to modify and maintain it. The cost is high, which in turn leads to a decrease in user experience.

Contents of the invention

Embodiments of the present disclosure at least provide an image processing method, device, device, medium, and program.

In a first aspect, an embodiment of the present disclosure provides an image processing method, the method is executed by an electronic device, including:

Get the image to be processed;

Based on the target task corresponding to the image to be processed, determine a plurality of computing node information and execution process information; wherein, the computing nodes corresponding to the plurality of computing node information are used to complete different preset functions;

Based on the plurality of computing node information and the execution process information, generate an execution process corresponding to the execution process information; wherein the execution process includes at least one computing node information;

Based on the execution flow, calculation graph information is generated, and the image to be processed is processed according to the calculation graph information to obtain an image processing result.

In this aspect, the post-processing algorithm can be preliminarily abstracted into a hierarchical structure, that is, computing nodes and execution processes, by using the execution process generated by abstracting the computing node information and execution process information corresponding to computing nodes with different preset functions, and then, The image to be processed is processed by using the calculation graph information composed of the execution flow, and the post-processing algorithm is further abstracted into a three-layer architecture system, namely, the calculation node (first layer), the execution flow (second layer) and the calculation graph (third layer). ). For the post-processing algorithm with hierarchical architecture, since the computing nodes are used to complete the preset function (such as a single function), the existing computing nodes are used to flexibly construct the post-processing calculation graph to meet the processing conditions of the target task , for different tasks, there is no need to repeatedly store code blocks with the same function (corresponding to the above-mentioned computing nodes), which reduces redundant code storage, reduces the storage load of the platform, improves the flexibility of using computing nodes, and reduces the Difficulty managing post-processing algorithms. In addition, the construction and adjustment of processing algorithms no longer require R&D personnel to adjust the code, but only need the staff to configure the information required for the calculation graph, and then construct the calculation graph based on the generated calculation graph information to process the image to be processed, which improves the image quality. processing efficiency.

In a second aspect, an embodiment of the present disclosure further provides an image processing device, including:

An image acquisition module configured to acquire images to be processed;

The information determination module is configured to determine a plurality of computing node information and execution process information based on the target task corresponding to the image to be processed; wherein, the computing nodes corresponding to the plurality of computing node information are used to complete different preset functions ;

A process generation module configured to generate an execution process corresponding to the execution process information based on the plurality of computing node information and the execution process information; wherein the execution process includes at least one computing node information;

The image processing module is configured to generate calculation graph information based on the execution flow, and process the image to be processed according to the calculation graph information to obtain an image processing result.

In a third aspect, an embodiment of the present disclosure further provides a computer device, including: a processor, a memory, and a bus, the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processing The processor communicates with the memory through a bus, and when the machine-readable instructions are executed by the processor, the above-mentioned first aspect, or the steps of any possible image processing method in the first aspect are executed.

In a fourth aspect, embodiments of the present disclosure further provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned first aspect, or any of the first aspects of the first aspect, may be executed. Steps of a possible image processing method.

An embodiment of the present disclosure also provides a computer program, where the computer program includes computer readable codes, and when the computer readable codes run in an electronic device, the processor of the electronic device executes any of the above embodiments The image processing method described.

For the effect description of the above image processing apparatus, computer equipment and storage medium, please refer to the description of the above image processing method, which will not be repeated here.

In order to make the above-mentioned objects, features and advantages of the present disclosure more comprehensible, preferred embodiments will be described in detail below together with the accompanying drawings.

Description of drawings

In order to illustrate the technical solutions of the embodiments of the present disclosure more clearly, the following will briefly introduce the accompanying drawings used in the embodiments. The accompanying drawings here are incorporated into the specification and constitute a part of the specification. The drawings show the embodiments consistent with the present disclosure, and are used together with the description to explain the technical solution of the present disclosure. It should be understood that the following drawings only show some embodiments of the present disclosure, and therefore should not be regarded as limiting the scope. For those skilled in the art, they can also make From these drawings other related drawings are obtained.

FIG. 1 shows a flowchart of an image processing method provided by an embodiment of the present disclosure;

FIG. 2 shows a schematic diagram of a system architecture applying an image processing method according to an embodiment of the present disclosure;

FIG. 3 shows a schematic diagram of determining dependency information between computing nodes based on input information and output information of each computing node provided by an embodiment of the present disclosure;

FIG. 4 shows a schematic diagram of a calculation graph provided by an embodiment of the present disclosure;

FIG. 5 shows an execution flowchart of an application scenario of an image processing method provided by an embodiment of the present disclosure;

FIG. 6 shows a schematic diagram of an image processing device provided by an embodiment of the present disclosure;

FIG. 7 shows a schematic structural diagram of a computer device provided by an embodiment of the present disclosure.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only It is a part of the embodiments of the present disclosure, but not all of them. The components of the disclosed embodiments generally described and illustrated in the figures herein may be arranged and designed in a variety of different configurations. Accordingly, the following detailed description of the embodiments of the present disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely represents selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative effort shall fall within the protection scope of the present disclosure.

In addition, the terms "first", "second" and the like in the description and claims in the embodiments of the present disclosure and the above drawings are used to distinguish similar objects, and not necessarily used to describe a specific order or sequence. It is to be understood that the terms so used are interchangeable under appropriate circumstances such that the embodiments described herein can be practiced in sequences other than those illustrated or described herein.

"Plural or several" mentioned herein means two or more. "And/or" describes the association relationship of associated objects, indicating that there may be three types of relationships, for example, A and/or B may indicate: A exists alone, A and B exist simultaneously, and B exists independently. The character "/" generally indicates that the contextual objects are an "or" relationship.

Multiple or multiple in the embodiments of the present disclosure may refer to at least two or at least two, respectively.

After research, it is found that for the post-processing algorithm of the AI remote sensing image interpretation model, when the traditional technology is applied to different tasks and the post-processing algorithm is temporarily adjusted, the R&D personnel need to modify the code of the post-processing algorithm. In related technologies, when post-processing codes are often modified, many versions of post-processing algorithms for different tasks are formed. Different versions of post-processing algorithms may include codes with repeated functions, resulting in support for AI remote sensing detection models and post-processing algorithms. The platform where the processing algorithm runs stores excessive redundant codes, resulting in a large storage load on the platform. At the same time, it is difficult to uniformly manage different versions of the post-processing algorithm. In addition, every time the user needs to adjust the post-processing algorithm, R&D personnel are required to modify and maintain it. The cost is high, which in turn leads to a decrease in user experience.

Based on the above research, the embodiment of the present disclosure provides an image processing method, which can preliminarily abstract the post-processing algorithm into a hierarchical architecture, that is, the computing nodes and the execution process, and then use the calculation graph information composed of the execution process to process the image to be processed, and further abstract the post-processing algorithm into a three-layer architecture system, namely, the computing node (first layer), the execution process (second layer) and computational graph (third layer). For the post-processing algorithm with hierarchical architecture, since the computing nodes are used to complete the preset function (such as a single function), the existing computing nodes are used to flexibly construct the post-processing calculation graph to meet the processing conditions of the target task , for different tasks, there is no need to repeatedly store code blocks with the same function (corresponding to the above-mentioned computing nodes), which reduces redundant code storage, reduces the storage load of the platform, improves the flexibility of using computing nodes, and reduces the Difficulty managing post-processing algorithms. In addition, the construction and adjustment of processing algorithms no longer require R&D personnel to adjust the code, but only need the staff to configure the information required for the calculation graph, and then construct the calculation graph based on the generated calculation graph information to process the image to be processed, which improves the image quality. processing efficiency.

The defects in the above solutions are all the results obtained by the inventor after practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed by the present disclosure below for the above problems should be the result of the inventor Contributions made to this disclosure during the course of this disclosure.

It should be noted that like numerals and letters denote similar items in the following figures, therefore, once an item is defined in one figure, it does not require further definition and explanation in subsequent figures.

The special nouns involved in the embodiments of the present disclosure are further explained below:

1. Post-processing: After the image is reasoned and recognized by the AI remote sensing detection model, it is then subjected to some processing such as splicing, fusion, edge blurring, and refinement, so that the accuracy of the results and the display effect are further improved.

2. Computation graph: an abstract data structure composed of computing nodes (operators) and lines (operator dependencies).

3. Remote sensing interpretation, recognition and analysis of remote sensing images.

In order to facilitate the understanding of this embodiment, an image processing method disclosed in the embodiment of the present disclosure is firstly introduced in detail. The image processing method provided in the embodiment of the present disclosure is generally executed by a computer device with certain computing capabilities. In some possible implementation manners, the image processing method may be implemented by a processor invoking computer-readable instructions stored in a memory.

The image processing method provided by the embodiment of the present disclosure will be described below by taking the execution subject as a computer device as an example.

In order to facilitate the understanding of this embodiment, the application scenario of an image processing method disclosed in the embodiment of the present disclosure is firstly introduced. The image processing method provided in the embodiment of the present invention can be applied to post-processing of image recognition in the field of remote sensing image interpretation , which is the post-processing for the recognition and analysis of remote sensing images. Exemplarily, the post-processing for the building extraction model may include functions such as removal of small image spots, regularization of building extraction results, and vectorization; the post-processing for the water body identification model may include removal of small patches, boundary blurring, boundary growth, Multi-model result fusion and other functions.

In the following, the image processing method provided by the embodiment of the present disclosure is used to meet the task requirements of different industry applications, and the post-processing algorithm of the remote sensing AI detection model is appropriately adjusted.

Referring to FIG. 1 , which is a flow chart of an image processing method provided by an embodiment of the present disclosure, the method is executed by an electronic device, and the method includes steps S101 to S104, wherein:

S101: Acquire images to be processed.

In this step, the image to be processed may include the category image that has been processed by the AI remote sensing detection model, that is, the result of remote sensing interpretation, and an image representing the category information of the AI remote sensing detection. The image to be processed is a result image that needs to be post-processed, and the result image may or may not be a remote sensing image.

S102: Based on the target task corresponding to the image to be processed, determine multiple computing node information and execution process information.

In some embodiments, the computing nodes corresponding to the multiple computing node information are used to implement different preset functions.

In this step, the target task may be a task of post-processing the image to be processed. The target tasks are different, and the target functions contained in the corresponding post-processing algorithms are different. Exemplarily, the target task includes a water body detection task. For example, the water body detection task may include pre-remote sensing interpretation processing and post-processing; wherein, the post-processing may be performed on the initial water body detection results, that is, remote sensing interpretation for images containing water bodies The result (that is, the first remote sensing image) is an algorithm for further processing; the target function included in the algorithm can be functions such as removing small patches, boundary blurring, boundary growth, and multi-model result fusion; or, exemplary, the target task Including the building detection task, for example, the building detection task may include pre-remote sensing interpretation processing and post-processing; where the post-processing may be for the initial building detection result, that is, the result of remote sensing interpretation for the image containing the building ( That is, the second remote sensing image), an algorithm for further processing; the target functions included in the algorithm can be functions such as removing small image spots, regularizing building extraction results, and vectorizing.

Here, the computing nodes corresponding to the multiple computing node information are used to implement different preset functions, wherein the preset function may include a single function, which is the minimum function among the target functions. For example, if the target function includes regularization of building extraction results, then the minimum function in the target function includes extraction function, result regularization function and so on. A compute node may include function code capable of performing a single predetermined function. The computing node information may include identification information of the computing node, such as name, serial number, and the like.

Taking the target task as an example of the post-processing task of removing small image spots, boundary blurring and vectorization of the image to be processed, based on the first subtask in the target task, that is, removing small image spots, determine at least one corresponding to the removal of small image spots Computing node information; based on the second subtask in the target task, that is, boundary blurring, determine at least one computing node information corresponding to boundary blurring; based on the third subtask in the target task, namely vectorization, determine at least one corresponding to vectorization Computing node information, here, can be the name of computing node A and computing node B corresponding to the removal of small image spots; the name of computing node C corresponding to boundary blur; the name of computing node D corresponding to vectorization, computing node The name of E and the name of computing node F, etc.

Here, each execution flow information corresponds to one execution flow, therefore, each execution flow information includes at least one computing node information included in its corresponding execution flow. Continuing the above example, the three subtasks correspond to three execution task information, that is, there are three execution processes correspondingly, where the first execution process information includes the computing node A information and computing node B information included in the corresponding first execution process ; The second execution process information includes the information of computing node C included in the corresponding second execution process; the third execution process information includes the information of computing node D included in the corresponding third execution process, computing node E information and computing node F information.

S103: Generate an execution process corresponding to the execution process information based on a plurality of computing node information and execution process information; wherein, the execution process includes at least one computing node information.

During specific implementation, the execution process corresponding to the execution process information can be generated according to at least one computing node information in the execution process information, and the dependency relationship information between each computing node corresponding to each computing node information, for example, the computing node A depends on computing node B, so it can be determined that the execution sequence of the execution process is to execute computing node B first, and then execute computing node A. Here, the execution process includes the name of computing node A, the name of computing node B, and the relationship between computing nodes. The sequence of execution does not include the function codes of computing node A and computing node B.

Here, the computing node information also includes input information and output information of the corresponding computing node. Determine the dependency relationship information between each computing node, for example, determine the dependency relationship information between multiple computing nodes based on the input information and output information of each computing node; based on the determined dependency relationship information and execution process information, An execution flow corresponding to the execution flow information is generated. Wherein, the input information/output parameter may include at least one of the following: the identification of the input parameter/output parameter of the calculation node, the type of the input parameter/output parameter; wherein, the type of the input parameter/output parameter, for example, may be an image, a digital Or parameter types such as pointers.

Exemplarily, the target task may include multiple subtasks, each subtask corresponds to at least one piece of execution flow information, and each piece of execution flow information corresponds to one execution flow, therefore, the target task may have multiple execution flows.

As shown in Figure 2, it is a schematic diagram of determining the dependency information between computing nodes based on the input information and output information of each computing node; wherein, the identification of the input parameter in the information of computing node A is a, and the identification of the output parameter is The identification is c; the identification of input parameters in the information of computing node B is b, and the identification of output parameters is d; the identification of input parameters in information of computing node C is c and d, and the identification of output parameters is e; information of computing node D The identifier of the input parameter in is d, the identifier of the output parameter is f; the identifiers of the input parameters in the computing node E information are e and f, and the identifier of the output parameter is g. Afterwards, when the input information and output information of each computing node are determined, since the identifier of the input parameter of computing node C is the identifier of the output parameter of computing node A, it can be determined that computing node C depends on computing node A, that is, Determine dependency information between computing node C and computing node A. Similarly, it can be determined that the dependency relationship information between computing nodes includes that computing node E depends on computing nodes C and computing nodes D, computing node C depends on computing nodes A and computing nodes B, and computing node D depends on computing nodes B.

Afterwards, based on the determined dependency information and execution flow information, the execution flow corresponding to the execution flow information is generated, which may include execution flow A, B→C, execution flow B→D, and execution flow C, D→E.

The above-mentioned input information and output information of computing nodes can be used to automatically determine the dependency information between computing nodes, and the execution process corresponding to the execution process information can be generated without specifying the execution sequence between computing nodes, which improves the processing of target tasks. efficiency.

S104: Generate calculation graph information based on the execution flow, and process the image to be processed according to the calculation graph information to obtain an image processing result.

The execution flow information may include the execution sequence of the corresponding execution flow. During specific implementation, when it is determined that there are multiple execution process information for the target task, calculation graph information can be generated based on the multiple execution processes and the execution order of each execution process; then, based on the execution order, the first execution process is determined; Input the image to be processed into the first computing node corresponding to the first execution process, and process it according to the first execution process and other execution processes corresponding to the execution order, until all the execution processes are executed, and the image processing result is obtained. Wherein, other execution processes include multiple execution processes corresponding to the target task except the first execution process.

Here, each execution process includes computing node information corresponding to at least one computing node to be executed by the execution process. The calculation graph information includes the calculation graph and the description information of the calculation graph. Wherein, the description information of the calculation graph may include information describing at least one execution process and at least one calculation node to be executed by each execution process, and the description information of the calculation graph is used to construct the calculation graph.

Exemplarily, the calculation graph can be constructed according to the description information of the calculation graph, and then each execution process in the calculation graph is run, and each calculation node is calculated in the execution process according to the execution order (that is, the execution interface of each calculation node is called, and the Calculation), so that the image processing can be completed in an orderly manner according to the execution order, and the image processing result can be obtained.

As shown in Figure 3, it is a schematic diagram of the calculation graph, including three execution processes, where the first execution process includes three computing nodes (S_1, S_2, and S_3, of which the first computing node is S_1) and their dependency information , that is, S_1→S_2→S_3, the second execution process includes two computing nodes (S_4 and S_5) and their dependency information, that is, S_4→S_5, and the third execution process includes two computing nodes (S_6 and S_7) and Its dependency information, namely S_6→S_7. Wherein, the execution sequence is to first execute S_1→S_2→S_3, then execute S_4→S_5, and finally execute S_6→S_7.

Here, the execution process is the minimum execution sequence of the calculation graph, which can complete at least one target function (such as the regularization of building extraction results), each calculation graph has at least one execution process, and the calculation node can complete the minimum preset function in the calculation graph (such as extract function).

Executing the target task, for example, as shown in FIG. 3 , includes three subtasks, that is, each subtask corresponds to an execution process. The interface of input parameters/output parameters of each computing node and the execution interface of the function code of the computing node can be respectively called according to the calculation graph, and the function code of the computing node can be called to execute the target task. In specific implementation, starting from the first computing node of the first execution process, call the function code of the computing node to process the input parameters, obtain the output result, store the output result through the output interface of the computing node, and wait for the input of the next computing node Called when parameters are entered. Exemplarily, call the function code of the first computing node S_1, and the input parameters, that is, the image to be processed, and use the function code of S_1 to process the image to be processed, store the output result in the memory 1, and then, from the memory In 1, call the input parameters of computing node S_2 (that is, the output result of S_1), and further process the output result of the previous computing node S_1, and store the processing result in memory 2, and cycle in turn until all executions are executed process to obtain image processing results.

Taking the target task including the water body detection task as an example, at this time, the image to be processed may include the first remote sensing image corresponding to the initial water body detection result obtained after the early remote sensing interpretation processing. Afterwards, the first remote sensing image is processed according to the execution flow in the calculation graph information, and the target water body detection result corresponding to the water body detection task is obtained.

During specific implementation, each execution process in the calculation graph can be run, the function code of each calculation node in the execution process can be called according to the execution order, and the post-processing task in the water body detection task can be executed to obtain the target water body detection result.

Taking the target task including building detection task as an example, at this time, the image to be processed may include the second remote sensing image corresponding to the initial building detection result obtained after the previous remote sensing interpretation processing. Afterwards, the second remote sensing image is processed according to the execution flow in the calculation graph information, and the target building detection result corresponding to the building detection task is obtained.

During specific implementation, each execution process in the calculation graph can be run, the function code of each calculation node in the execution process can be called according to the execution order, and the post-processing task in the building detection task can be executed to obtain the detection result of the target building.

In the embodiment of the present disclosure, the post-processing algorithm can be preliminarily abstracted into a hierarchical architecture by using the computing node information corresponding to the computing node abstracted as a single preset function and the execution process information to generate the execution process, that is, the computing node and the execution process , after that, use the calculation graph information composed of the execution process to process the image to be processed, and further abstract the post-processing algorithm into a three-layer architecture system, namely, the calculation node (the first layer), the execution process (the second layer) and the calculation graph ( the third floor). For the post-processing algorithm with a hierarchical structure, since the computing nodes are used to complete a single preset function, the existing computing nodes are used to flexibly construct the post-processing calculation graph to meet the processing conditions of the target task, and different tasks do not use Repeated storage of code blocks with the same function (corresponding to the above-mentioned computing nodes) reduces redundant redundant codes, reduces platform storage load, improves the flexibility of computing nodes, and reduces the management of post-processing algorithms difficulty. In addition, the construction and adjustment of processing algorithms no longer require R&D personnel to adjust the code, but only need the staff to configure the information required for the calculation graph, and then construct the calculation graph based on the generated calculation graph information to process the image to be processed, which improves the image quality. processing efficiency.

FIG. 4 shows a schematic diagram of a system architecture applying an image processing method according to an embodiment of the present disclosure; as shown in FIG. 4 , the system architecture includes: an image acquisition terminal 201 , a network 202 and a control terminal 203 . In order to support an exemplary application, the acquisition terminal 201 and the control terminal 203 establish a communication connection through the network 202, the image acquisition terminal 201 reports the image to be processed to the control terminal 203 through the network 202, and when the control terminal 203 acquires the image to be processed , based on the target task corresponding to the image to be processed, determine a plurality of computing node information and execution process information; wherein, the computing nodes corresponding to the plurality of computing node information are used to complete different preset functions; based on the multiple computing node information and the execution process information, generating an execution process corresponding to the execution process information; wherein, the execution process includes at least one computing node information; based on the execution process, generating calculation graph information, and according to the The calculation graph information processes the image to be processed to obtain an image processing result. Finally, the control terminal 203 uploads the image processing result to the network 202 and sends it to the image acquisition terminal 201 through the network 202 .

As an example, the image acquisition terminal 201 may include an image acquisition device, and the control terminal 203 may include a vision processing device capable of processing visual information or a remote server. The network 202 may be connected in a wired or wireless manner. Wherein, when the control terminal 203 is a visual processing device, the image acquisition terminal 201 can communicate with the visual processing device through a wired connection, such as performing data communication through a bus; when the control terminal 203 is a remote server, the current scene image acquisition terminal 201 can perform data interaction with a remote server through a wireless network.

Or, in some scenarios, the image acquisition terminal 201 may be a vision processing device with a video capture module, or a host with a camera. At this time, the display method in the augmented reality scene of the embodiment of the present disclosure may be executed by the image acquisition terminal 201 , and the above-mentioned system architecture may not include the network 202 and the control terminal 203 .

In some embodiments, if the computing nodes corresponding to the plurality of computing node information determined in S102 include undefined computing nodes, the undefined computing nodes need to be generated during the process of building the computing graph.

During specific implementation, when it is determined that the computing nodes corresponding to the multiple computing node information include undefined computing nodes, the corresponding computing nodes are generated based on the computing node information of the undefined computing nodes; call The computing node corresponding to each computing node information processes the image to be processed to obtain an image processing result.

An undefined computing node may be an algorithm that has not been implemented in the software, for example, the function code of the computing node is not added. Defining a compute node may be adding the algorithm (function code) to the software.

Exemplarily, an undefined computing node may have already stored the function code of the computing node, but has not been configured with an execution interface for calling the function code of the computing node. For example, when it is determined that the undefined computing node is not In the case where the execution interface of the function code corresponding to the extraction function and the interface of the function code corresponding to the result regularization function are configured to call the building extraction result regularization target function, the user can configure the function code corresponding to the extraction function. An execution interface (configurable through a configuration file), and an execution interface for configuring the function code corresponding to the call result regularization function. Afterwards, construct a calculation graph (including newly defined computing nodes) according to the description information of the computing graph, and then run each execution process in the calculation graph, and calculate each computing node in the execution process (that is, call the execution interface of each computing node , to complete the calculation), until the image processing result is obtained.

Exemplarily, an undefined computing node may also be that the function code of the computing node is not stored, and the execution interface for calling the function code of the computing node is not configured. For example, the function code of building extraction result regularization can be determined from the post-processing algorithm of the historical task, and the function code corresponding to the execution extraction function and the function code corresponding to the result regularization function can be determined from it; after that, the extraction function is configured and called The execution interface of the corresponding function code (which can be configured through the configuration file), and the execution interface of the function code corresponding to the call result regularization function, can realize the definition of the undefined computing node and generate the corresponding calculate node.

As mentioned above, there may be undefined computing nodes in the computing nodes corresponding to the computing node information. Therefore, it is necessary to generate corresponding computing nodes based on the computing node information of the undefined computing nodes, so as to avoid subsequent task execution based on the computing graph. Image processing is interrupted due to calling undefined computing nodes, thereby improving the efficiency of image processing.

In some embodiments, when the input information/output information of a certain execution process occupies a large amount of memory, the input information of the execution process can also be processed in batches to obtain processing result information with a smaller memory, and batch processing The execution flow needs to be executed cyclically. In specific implementation, when the execution flow includes a cyclic execution flow, the output information of the previous execution flow of the cyclic execution flow is obtained, and the output information is divided into multiple sub-information; the multiple sub-information are respectively Enter the cyclic execution process for processing, and obtain the processing result information corresponding to each sub-information; determine the target result information based on the processing result information corresponding to each sub-information, and input the target result information into the next execution process of the cyclic execution process until execution After completing all the execution processes, the image processing result is obtained.

Here, the cyclic execution process is a cyclic execution process, and the number of cyclic executions can be set according to the memory size occupied by the input information/output information of the cyclic execution process. The output information of the previous execution process of the cyclic execution process is the input information of the cyclic execution process, and the input information is divided into multiple sub-information. Processing the sub-information occupying a small memory can obtain the processing result information (that is, part of the information in the output information of the cyclic execution process) occupying a small memory. Exemplarily, it is known that the preset maximum memory is the third-generation mobile communication technology (3rd-Generation, 3G). In the case that the memory occupied by the input information of the cyclic execution flow is 3G, since the processing process of the cyclic execution flow also includes There is a lot of intermediate information, so the memory occupied by the output information of the cyclic execution process is larger than 3G. Therefore, the output information of the previous execution process can be divided into 2 sub-information, and the number of cyclic executions of the cyclic execution process is 2 times. If the performance impact of device operation is also considered, the output information of the previous execution process can be divided into 3 sub-information, and the number of cyclic executions of the cyclic execution process is 3 times.

Segmentation, for example, in the case where the memory occupied by the output information of the previous execution process is greater than the preset value, the output information of the previous execution process may be divided according to the memory occupied by each sub-information is less than the preset maximum memory amount , divided into multiple sub-information. Exemplarily, when the memory occupied by the output information of the previous execution process is less than the preset value, and the memory occupied by the processing result is greater than the preset value, the processing result is divided so that the memory occupied by it is less than the preset maximum memory amount, Afterwards, according to the number of divisions, the output information of the previous execution process is evenly divided into multiple sub-information.

Afterwards, in one case, based on the association relationship between multiple sub-information obtained by segmentation, the associated two sub-information can be used as adjacent input sub-information to determine the input sequence during loop execution; after that, according to the processed The output sequence continuously integrates the processing result information corresponding to each sub-information to obtain the target processing result. In another case, input multiple sub-information out of order, for example, input multiple sub-information respectively into the loop execution process for processing, you can get the processing result information corresponding to each sub-information, after that, process the processing result information corresponding to each sub-information After the final integration, the integrated target processing results can be obtained. In different application scenarios, the data integration process is different, which is not specifically limited in the embodiments of the present disclosure.

Here, since the amount of information data to be processed is relatively large, the sub-information after the batch processing information of the cyclic execution flow can be used to fuse the processing result information corresponding to each sub-information obtained by the cyclic execution flow processing, and it can be obtained by using Target result information representing the processing result of the entire information, so that the process can continue to be executed. In this embodiment, the sub-information is processed in batches by means of a cyclic execution flow, which can reduce the computer resources invoked by each processing process, thereby reducing equipment loss.

In some embodiments, the user can pre-configure the configuration file corresponding to the target task, and then download the configuration file to the computer device to prepare for executing the post-processing function corresponding to the target task.

Here, the configuration file includes a plurality of computing node information, at least one execution process information, and execution interface information of each computing node called when running the computing graph, and the configuration file is used to configure information for generating the computing graph information.

For S102, based on the target task corresponding to the image to be processed, determine multiple computing node information and at least one execution process information, which can be achieved through the following process: first, obtain the configuration file corresponding to the target task; second, obtain the target task from the configuration file Corresponding pieces of computing node information and at least one execution process information.

Exemplarily, for the target task, at least one execution process information can be obtained from the configuration file, and then at least one execution process item is created, and at least one computing node information corresponding to the execution process item is added to each execution process item, according to The dependency information between computing nodes corresponding to each computing node information generates an execution process.

Exemplarily, the target task is to perform boundary blurring and post-processing for removing small plaques. For example, the configuration file includes computing node information Q_1 corresponding to the boundary blur function, computing node information Q_2 and Q_3 corresponding to removing small plaques, execution process information P_1 is post-processing of boundary blurring, and execution process information P_2 is post-processing of removing small plaques. Afterwards, according to the execution process information P_1 in the configuration file, determine the execution process P_1^' as the computing node corresponding to the execution computing node information Q_1; according to the execution process information P_2 in the configuration file and the dependency information between Q_2 and Q_3, determine The execution process P_2^' is to first execute the computing node corresponding to the computing node information Q_2, and then execute the computing node corresponding to the computing node information Q_3. Afterwards, the calculation graph information is generated according to at least one execution process, and the function code corresponding to the computing node can be called according to the calculation graph information to execute the post-processing algorithm, complete the post-processing process of the image to be processed, and obtain the image processing result.

Here, using the configuration file to store multiple computing node information and at least one execution process information can ensure that the above-mentioned computing node information and execution process information used to generate the calculation graph information can be quickly obtained according to the configuration file, and the calculation graph information is improved. Efficiency and accuracy of information generation.

In some embodiments, the standard detection result of the object to be detected in the target task stored in the history can also be obtained; according to the calculation graph information and the standard detection result, the image to be processed is processed to obtain the image processing result. It uses the pre-set task threshold information to obtain accurate image processing results.

Exemplarily, taking the target task as a water body detection task as an example, that is, the object to be detected is a water body, the historically stored standard detection results for water bodies can be obtained, and then the first remote sensing image is processed according to the calculation map information to obtain the water body The detection result of the target water body corresponding to the detection task, and then compare the detection result of the target water body with the standard detection result for the water body stored in history. detection.

Here, different application scenarios have different task threshold information, and the embodiment of the present disclosure does not specifically limit it, without departing from the scope of the present disclosure. Those skilled in the art can make various substitutions and modifications for the task threshold information, and these substitutions and modifications are all should fall within the scope of this disclosure. Since the task threshold information can be flexibly set according to specific application scenarios, the use of the pre-set task threshold information to determine the image processing result can improve the practicability of the image processing method and the accuracy of the image processing result.

As shown in FIG. 5, it is an execution flowchart of an application scenario of an image processing method provided by an embodiment of the present disclosure, including:

S401: Create post-processing calculation graph items.

The calculation graph item is used to receive the execution process and calculation node information, and generate calculation graph information.

S402: Add an execution flow item.

The purpose of adding an execution process is to establish an execution process. To add an execution process item, for example, at least one execution process item may be added according to the execution process information in the configuration file. Execution process items are used to receive compute node information.

S403: Add computing node information for each execution process item.

S404: Determine whether the computing node corresponding to the computing node information in each execution process is defined; if it is defined, execute S405; if not, execute S408.

S405: Determine the execution process corresponding to each execution process item according to the computing node information added to each execution process item and the dependency relationship information between computing nodes corresponding to each computing node information.

After the computing node information is added, the execution process item is completed, and then, the execution process is generated according to the dependency relationship information between computing nodes corresponding to each computing node information. The execution process can call the computing node corresponding to the computing node information according to its internal computing node information.

S406: Generate calculation graph information according to at least one execution flow;

S407: Call the function code corresponding to the computing node according to the computing graph information, execute the post-processing algorithm, complete the post-processing process of the image to be processed, obtain the image processing result, and end the process.

S408: Define an undefined computing node.

In the embodiment of the present disclosure, by decomposing the remote sensing interpretation post-processing algorithm into many computing nodes, the simplification of the preset functions that the computing nodes can complete is maintained. When the post-processing algorithm needs to be adjusted, there is no need to modify the existing The preset function does not need to modify the code, just change the combination of calculation node information in the post-processing calculation graph information to meet different post-processing requirements.

Those skilled in the art can understand that in the above method of specific implementation, the writing order of each step does not mean a strict execution order and constitutes any limitation on the implementation process. The specific execution order of each step should be based on its function and possible The inner logic is OK.

Based on the same inventive concept, the embodiment of the present disclosure also provides an image processing device corresponding to the image processing method. Since the problem-solving principle of the device in the embodiment of the present disclosure is similar to the above-mentioned image processing method in the embodiment of the present disclosure, the implementation of the device Reference can be made to the implementation of the method, and repeated descriptions will not be repeated.

As shown in FIG. 6 , it is a schematic diagram of an image processing device provided by an embodiment of the present disclosure, and the device includes: an image acquisition module 501, an information determination module 502, a flow generation module 503, and an image processing module 504; wherein,

An image acquisition module 501 configured to acquire an image to be processed;

The information determination module 502 is configured to determine a plurality of computing node information and execution process information based on the target task corresponding to the image to be processed; wherein, the computing nodes corresponding to the plurality of computing node information are used to complete different presets Function;

The process generation module 503 is configured to generate an execution process corresponding to the execution process information based on the plurality of computing node information and the execution process information; wherein the execution process includes at least one computing node information;

The image processing module 504 is configured to generate calculation graph information based on the execution process, and process the image to be processed according to the calculation graph information to obtain an image processing result.

In some embodiments, the computing node information includes input information and output information of corresponding computing nodes;

The process generating module 503 is configured to determine dependency information among multiple computing nodes based on the input information and output information of each computing node; and generate the determined dependency information and the execution process information based on the determined Execution process corresponding to the execution process information described above.

In some embodiments, the device further includes a node generating module 505 configured to determine a plurality of computing node information based on the target task corresponding to the image to be processed, and determine the computing nodes corresponding to the multiple computing node information When the nodes include undefined computing nodes, generating corresponding computing nodes based on computing node information of the undefined computing nodes;

The image processing module 504 is configured to call a computing node corresponding to each computing node information to process the image to be processed according to the computing graph information, and obtain an image processing result.

In some embodiments, the execution process information includes the execution order of the corresponding execution process;

The image processing module 504 is configured to generate calculation graph information based on the multiple execution processes and the execution order of each execution process when it is determined that the target task has multiple execution process information; The above execution order is used to determine the first execution process; the image to be processed is input to the first computing node corresponding to the first execution process, and the process is performed according to the first execution process and other execution processes corresponding to the execution order processing, until all the execution processes are executed, and the image processing result is obtained.

In some embodiments, the execution process includes a cyclic execution process;

The image processing module 504 is configured to obtain the output information of the previous execution flow of the cyclic execution flow, and divide the output information into a plurality of sub-information; respectively input the plurality of sub-information into the cyclic execution flow to perform processing, respectively obtaining the processing result information corresponding to each sub-information; determining the target result information based on the processing result information corresponding to each sub-information, and inputting the target result information into the next execution process of the cyclic execution process until the execution is completed All execution processes are performed to obtain the image processing results.

In some embodiments, the information determining module 502 is configured to acquire a configuration file corresponding to the target task; and acquire multiple computing node information and at least one execution process information corresponding to the target task from the configuration file.

In some embodiments, the image processing module 504 is configured to obtain a standard detection result of the object to be detected in the target task stored in history; perform calculation on the image to be processed according to the calculation graph information and the standard detection result processing to get the image processing result.

In some embodiments, the image to be processed includes the first remote sensing image corresponding to the initial water body detection result; the target task includes a water body detection task;

The image processing module 504 is configured to process the first remote sensing image according to the execution flow in the calculation graph information, and obtain the target water body detection result corresponding to the water body detection task.

In some embodiments, the image to be processed includes a second remote sensing image corresponding to the initial building detection result; the target task includes a building detection task;

The image processing module 504 is configured to process the second remote sensing image according to the execution flow in the calculation graph information to obtain the target building detection result corresponding to the building detection task.

For the description of the processing flow of each module in the image processing device and the interaction flow between the modules, reference may be made to the relevant description in the above embodiment of the image processing method, which will not be described in detail here.

Based on the same technical idea, the embodiment of the present disclosure also provides a computer device. Referring to FIG. 7 , it is a schematic structural diagram of a computer device provided by an embodiment of the present disclosure, including:

processor 61 , memory 62 and bus 63 . Wherein, the memory 62 stores machine-readable instructions executable by the processor 61, and the processor 61 is used to execute the machine-readable instructions stored in the memory 62. When the machine-readable instructions are executed by the processor 61, the processor 61 executes The following steps: S101: Obtain the image to be processed; S102: Based on the target task corresponding to the image to be processed, determine the information of multiple computing nodes and the execution process information; wherein, the computing nodes corresponding to the multiple computing node information are used to complete different Preset function; S103: Based on multiple computing node information and execution process information, generate an execution process corresponding to the execution process information; wherein, the execution process includes at least one computing node information; S104: Based on the execution process, generate calculation graph information, and follow The calculation graph information processes the image to be processed to obtain the image processing result.

Above-mentioned memory 62 comprises memory 621 and external memory 622; Memory 621 here is also called internal memory, is used for temporarily storing the operation data in processor 61, and the data exchanged with external memory 622 such as hard disk, processor 61 communicates with memory 621 through memory 621. The external memory 622 performs data exchange. When the computer device is running, the processor 61 communicates with the memory 62 through the bus 63, so that the processor 61 executes the execution instructions mentioned in the above method embodiments.

Embodiments of the present disclosure further provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is run by a processor, the steps of the image processing method described in the foregoing method embodiments are executed. Wherein, the storage medium may be a volatile or non-volatile computer-readable storage medium.

An embodiment of the present disclosure also provides another computer program product, including computer instructions, and when the computer instructions are executed by a processor, the steps of the above-mentioned image processing method are implemented. Wherein, the computer program product may be any product capable of realizing the above-mentioned image processing method, and part or all of the solutions in the computer program product that contribute to the prior art may be implemented as a software product (such as a software development kit (Software Development Kit, SDK) ), the software product can be stored in a storage medium, and the computer instructions contained therein make relevant devices or processors execute some or all of the steps of the above-mentioned image processing method.

The device involved in the embodiments of the present disclosure may be at least one of a system, a method, and a computer program product. A computer program product may include a computer readable storage medium having computer readable program instructions thereon for causing a processor to implement various aspects of the present disclosure.

A computer readable storage medium may be a tangible device that can retain and store instructions for use by an instruction execution device. A computer readable storage medium may be, for example, but is not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. Examples of computer-readable storage media (a non-exhaustive list) include: portable computer disks, hard disks, Random Access Memory (RAM), Read-Only Memory (ROM), erasable Electrical Programmable Read Only Memory (EPROM) or flash memory, Static Random-Access Memory (Static Random-Access Memory, SRAM), Portable Compact Disc Read-Only Memory (CD-ROM), Digital Video Discs (DVDs), memory sticks, floppy disks, mechanically encoded devices such as punched cards or raised structures in grooves with instructions stored thereon, and any suitable combination of the foregoing. As used herein, computer-readable storage media are not to be construed as transient signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., pulses of light through fiber optic cables), or transmitted electrical signals.

The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to a respective computing/processing device, or to an external computer or external storage device over at least one of a network, such as the Internet, a local area network, a wide area network, and a wireless network. . The network may include at least one of copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and edge servers. A network adapter card or a network interface in each computing/processing device receives computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in each computing/processing device .

Computer program instructions for performing the operations of the present disclosure may be assembly instructions, Industry Standard Architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or in one or more source or object code written in any combination of programming languages, including object-oriented programming languages—such as Smalltalk, C++, etc., and conventional procedural programming languages, such as the “C” language or similar programming languages. Computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server implement. In cases involving a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or it may be connected to an external computer (for example, using Internet Service Provider to connect via the Internet). In some embodiments, electronic circuits, such as programmable logic circuits, FPGAs, or programmable logic arrays (Programmable Logic Arrays, PLAs), can be customized by using state information of computer-readable program instructions, which can execute computer-readable Read program instructions, thereby implementing various aspects of the present disclosure.

Those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the device described above can refer to the corresponding process in the foregoing method embodiment, and details are not repeated here. In the several embodiments provided in the present disclosure, it should be understood that the disclosed devices and methods may be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods. For example, multiple modules or components can be combined. Or some features can be ignored, or not implemented. In another point, the mutual coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces, and the indirect coupling or communication connection of devices or modules may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional module in each embodiment of the present disclosure may be integrated into one processing module, each module may exist separately physically, or two or more modules may be integrated into one module.

If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on this understanding, the technical solution of the present disclosure is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in various embodiments of the present disclosure. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disc and other media that can store program codes. .

Finally, it should be noted that: the above-mentioned embodiments are only specific implementations of the present disclosure, and are used to illustrate the technical solutions of the present disclosure, rather than limit them, and the protection scope of the present disclosure is not limited thereto, although referring to the aforementioned The embodiments have described the present disclosure in detail, and those skilled in the art should understand that any person familiar with the technical field can still modify the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present disclosure Changes can be easily imagined, or equivalent replacements can be made to some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in this disclosure. within the scope of protection. Therefore, the protection scope of the present disclosure should be defined by the protection scope of the claims.

Industrial Applicability

Embodiments of the present disclosure provide an image processing method, device, device, medium, and program, wherein the method includes: acquiring an image to be processed; determining multiple computing node information and executing Process information; wherein, the computing nodes corresponding to the multiple computing node information are used to complete different preset functions; based on the multiple computing node information and the execution process information, the execution process information corresponding to the execution process is generated A process; wherein, the execution process includes at least one computing node information; based on the execution process, calculation graph information is generated, and the image to be processed is processed according to the calculation graph information to obtain an image processing result.

Claims

An image processing method, the method being executed by an electronic device, comprising:

Get the image to be processed;

Based on the target task corresponding to the image to be processed, determine a plurality of computing node information and execution process information; wherein, the computing nodes corresponding to the plurality of computing node information are used to complete different preset functions;

Based on the plurality of computing node information and the execution process information, generate an execution process corresponding to the execution process information; wherein the execution process includes at least one computing node information;

Based on the execution flow, calculation graph information is generated, and the image to be processed is processed according to the calculation graph information to obtain an image processing result.
The method according to claim 1, wherein the computing node information includes input information and output information of a corresponding computing node;

The generating the execution flow corresponding to the execution flow information based on the multiple computing node information and the execution flow information includes:

Determine dependency information among multiple computing nodes based on input information and output information of each computing node;

Based on the determined dependency relationship information and the execution flow information, an execution flow corresponding to the execution flow information is generated.
The method according to claim 1 or 2, wherein, after determining a plurality of computing node information based on the target task corresponding to the image to be processed, further comprising:

When it is determined that the computing nodes corresponding to the plurality of computing node information include undefined computing nodes, generating corresponding computing nodes based on the computing node information of the undefined computing nodes;

The processing of the image to be processed according to the calculation graph information to obtain an image processing result includes:

The computing node corresponding to each computing node information is invoked according to the computing graph information to process the image to be processed to obtain an image processing result.
The method according to any one of claims 1 to 3, wherein the execution process information includes the execution order of the corresponding execution process;

The calculation graph information is generated based on the execution process, including:

When it is determined that there are multiple execution process information for the target task, generating calculation graph information based on the multiple execution processes and the execution order of each execution process;

The processing of the image to be processed according to the calculation graph information to obtain an image processing result includes:

Determining a first execution flow based on the execution sequence;

Input the image to be processed into the first computing node corresponding to the first execution process, and process it according to the first execution process and other execution processes corresponding to the execution order until all the execution processes are executed, Obtain the image processing result.
The method according to claim 4, wherein the execution process comprises a cyclic execution process;

The processing of the image to be processed according to the description to obtain an image processing result includes:

Obtaining the output information of the previous execution process of the cyclic execution process, and dividing the output information into a plurality of sub-information;

Input the plurality of sub-information respectively into the loop execution flow for processing, and respectively obtain the processing result information corresponding to each sub-information;

Determine the target result information based on the processing result information corresponding to each sub-information, and input the target result information into the next execution process of the cyclic execution process until all the execution processes are executed to obtain the image processing result.
The method according to any one of claims 1 to 5, wherein said determining a plurality of computing node information and at least one execution process information based on the target task corresponding to the image to be processed includes:

Obtain a configuration file corresponding to the target task;

Acquiring multiple computing node information and at least one execution process information corresponding to the target task from the configuration file.
The method according to any one of claims 1 to 6, wherein, the processing of the image to be processed according to the calculation graph information to obtain an image processing result includes:

Obtaining the standard detection results of the object to be detected in the target task stored in history;

According to the calculation graph information and the standard detection result, the image to be processed is processed to obtain an image processing result.
The method according to any one of claims 1 to 7, wherein the image to be processed includes at least one of the first remote sensing image corresponding to the initial water body detection result and the second remote sensing image corresponding to the initial building detection result ;

The processing of the image to be processed according to the calculation graph information to obtain an image processing result includes at least one of the following:

Process the first remote sensing image according to the execution flow in the calculation graph information, and obtain the target water body detection result corresponding to the water body detection task in the target task;

The second remote sensing image is processed according to the execution flow in the calculation graph information to obtain a target building detection result corresponding to the building detection task in the target task.
An image processing device, comprising:

An image acquisition module configured to acquire images to be processed;

The information determination module is configured to determine a plurality of computing node information and execution process information based on the target task corresponding to the image to be processed; wherein, the computing nodes corresponding to the plurality of computing node information are used to complete different preset functions ;

A process generation module configured to generate an execution process corresponding to the execution process information based on the plurality of computing node information and the execution process information; wherein the execution process includes at least one computing node information;

The image processing module is configured to generate calculation graph information based on the execution flow, and process the image to be processed according to the calculation graph information to obtain an image processing result.
A computer device, comprising: a processor, a memory, and a bus, the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor communicates with the memory through the bus , when the machine-readable instructions are executed by the processor, the steps of the image processing method according to any one of claims 1 to 8 are executed.
A computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the image processing method according to any one of claims 1 to 8 are executed.
A computer program, the computer program comprising computer readable code, in the case of the computer readable code running in an electronic device, the processor of the electronic device executes to implement any one of claims 1 to 8 The image processing method described.