Workflow-based remote sensing image data automatic processing method and device

Abstract

The invention provides a workflow-based remote sensing image data automatic processing method and device, and belongs to the technical field of remote sensing image data processing, and the method comprises the steps: publishing a remote sensing image data processing algorithm as an available operator node in a workflow; creating a custom workflow and storing the custom workflow by utilizing the released operator nodes; creating an automatic production line by using the self-defined workflow, and providing data required by algorithm operation for operator nodes used in the production line; and constructing a process engine, and executing the operator nodes in sequence according to the sequence of the operator nodes configured in the production line so as to automatically process the remote sensing image data. According to the remote sensing image data automatic processing method based on the workflow, the processing algorithm can be published as an operator node, the self-defined workflow is built according to different requirements, automatic processing of the remote sensing image data is achieved, the processing efficiency of the remote sensing image data is remarkably improved, the operation difficulty is reduced, and the method has wide applicability.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image data processing, and in particular to a remote sensing image data automatic processing method and device based on workflow. Background Art

[0002] With the advent of the era of global observation, satellite remote sensing information is playing a significant role in promoting the development of human society in all fields. With the continuous improvement of the resolution of remote sensing satellite sensors, the remote sensing images generated have also increased dramatically, and they have become increasingly important due to their ability to quickly obtain high-resolution images. The rapid processing and production of massive remote sensing data has become an increasingly unstoppable demand, while the existing software resources in the field of automated and process-based processing of remote sensing data are still relatively scarce. Summary of the invention

[0003] In view of this, the present invention provides a workflow-based remote sensing image data automatic processing method and device, which are used to solve the problem that remote sensing image data cannot be processed in large quantities and automatically at present.

[0004] In order to solve the above technical problems, in a first aspect, the present invention provides a remote sensing image data automatic processing method based on a workflow, the method comprising: Publish the processing algorithms for remote sensing image data as operator nodes available in the workflow; Use the published operator nodes to create and save custom workflows; Using the custom workflow, a production line for automatically processing remote sensing image data is created, and the operator nodes used in the production line are provided with data required for algorithm operation; A process engine is constructed to provide automated operation capability for the production line, and each operator node is executed in sequence according to the order of the operator nodes configured in the production line to automatically process the remote sensing image data to obtain a processing result.

[0005] Optionally, the step of publishing the remote sensing image data processing algorithm as an operator node available in the workflow includes: Upload the image used by the processing algorithm to the image warehouse, and enter the image address and image description information into the publishing parameters; Entering relevant information of the processing algorithm into the publishing parameters, the relevant information including the environment required for the processing algorithm to run, the open port of the algorithm service, the access address of the algorithm service, the storage path of the algorithm code, and the transmission parameter information of the algorithm; According to the publishing parameters, the processing algorithm is published as an operator node available in the workflow.

[0006] Optionally, the step of using the published operator nodes to create and save a custom workflow includes: Define a target data structure for storing a custom workflow, wherein the target data structure includes description information of the custom workflow, operator nodes included in the custom workflow, connection relationships between operator nodes, and data transfer relationships between operator nodes having connection relationships; Select a target operator node from the published operator nodes, and create a custom workflow according to the target operator node; Serialize the created custom workflow into the target data structure.

[0007] Optionally, the step of providing the operator nodes used in the production line with data required for algorithm operation includes: Calculating the remaining parameters of each operator node used in the production line, where the remaining parameters are other input parameters required by the operator node in addition to the data transmitted by the previous operator node; Complete the remaining parameters.

[0008] Optionally, the step of building a process engine to provide automated operation capability for the production line includes: Create a production line startup node class, the startup node class is used to control the startup of the production line; Creating an algorithm node class, wherein the algorithm node class is used to dynamically create the operator nodes required by the production line, and to create an algorithm container according to the publishing information of the operator nodes and complete the algorithm running process; Creating an end node class, wherein the end node class is used to control the end of the production line; A process control class is created, where the process control class is used to control the execution order of each operator node in the production line according to the process relationship of the production line.

[0009] In a second aspect, the present invention further provides a workflow-based remote sensing image data automatic processing device, the device comprising: The publishing module is used to publish the processing algorithms of remote sensing image data as operator nodes available in the workflow; The workflow creation module is used to create and save custom workflows using published operator nodes; A production line creation module, used to create a production line for automatically processing remote sensing image data using the custom workflow, and provide the operator nodes used in the production line with data required for algorithm operation; The automated processing module is used to construct a process engine to provide automated operation capabilities for the production line, and to execute each operator node in sequence according to the order of the operator nodes configured in the production line to automatically process the remote sensing image data to obtain a processing result.

[0010] Optionally, the publishing module includes: A first input unit is used to upload the image used by the processing algorithm to the image warehouse, and input the image address and image description information into the publishing parameters; A second input unit is used to input relevant information of the processing algorithm into the publishing parameters, wherein the relevant information includes the environment required for the operation of the processing algorithm, the open port of the algorithm service, the access address of the algorithm service, the storage path of the algorithm code, and the transmission parameter information of the algorithm; A publishing unit is used to publish the processing algorithm as an operator node available in the workflow according to the publishing parameters.

[0011] Optionally, the workflow creation module includes: A data structure definition unit, used to define a target data structure for storing a custom workflow, wherein the target data structure includes description information of the custom workflow, operator nodes included in the custom workflow, connection relationships between operator nodes, and data transfer relationships between operator nodes having connection relationships; A workflow creation unit, used to select a target operator node from the published operator nodes, and create a custom workflow according to the target operator node; A serialization unit is used to serialize the created custom workflow into the target data structure.

[0012] Optionally, the production line creation module includes: A calculation unit, used to calculate the remaining parameters of each operator node used in the production line, wherein the remaining parameters are other input parameters required by the operator node in addition to the data transmitted by the previous operator node; The completion unit is used to complete the remaining parameters.

[0013] Optionally, the production line creation module includes: A startup node class creation unit, used to create a production line startup node class, wherein the startup node class is used to control the startup of the production line; An algorithm node class creation unit, used to create an algorithm node class, the algorithm node class is used to dynamically create the operator nodes required by the production line, and is used to create an algorithm container according to the release information of the operator node and complete the algorithm running process; An end node class creation unit, used to create an end node class, wherein the end node class is used to control the end of the production line; The process control class creation unit is used to create a process control class, and the process control class is used to control the execution order of each operator node in the production line according to the process relationship of the production line.

[0014] The beneficial effects of the above technical solution of the present invention are as follows: In the embodiment of the present invention, by publishing the processing algorithm of remote sensing image data as an operator node and building a custom workflow according to different needs, the automatic processing of remote sensing image data is realized, which significantly improves the processing efficiency of remote sensing image data, reduces the difficulty of operation, and has wide applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flowchart of a workflow-based automated remote sensing image data processing method in Embodiment 1 of the present invention; Figure 2 It is a schematic diagram of the structure of a workflow-based automatic processing device for remote sensing image data in the second embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0017] With the advent of the era of global observation, satellite remote sensing information is playing a significant role in promoting all areas of human social development. With the continuous improvement of the resolution of remote sensing satellite sensors, the remote sensing images generated have also increased dramatically, and their ability to quickly obtain high-resolution images has occupied an increasingly important position. The rapid processing and production of massive remote sensing data has become an increasingly unstoppable demand, while the existing software resources in the field of automated and process-based processing of remote sensing data are still relatively scarce. Based on the increasing demand for efficient processing and automated monitoring of image data for product production, the present invention proposes a method for customizing the configuration of a workflow process, and can automatically generate remote sensing data processing results based on a customized workflow process. Compared with traditional remote sensing data processing methods, this method has the advantages of automated processing, simple operation, and wide applicability, which significantly improves the efficiency of remote sensing image data processing.

[0018] Therefore, please refer to Figure 1 , Figure 1A flowchart of a workflow-based remote sensing image data automatic processing method provided in Embodiment 1 of the present invention, the method comprising the following steps: Step 11: Publish the remote sensing image data processing algorithm as an operator node available in the workflow.

[0019] In this step, the processing algorithms for remote sensing image data are existing algorithms, including preprocessing algorithms, feature extraction algorithms, classification and target extraction algorithms, change monitoring algorithms, etc. Specifically, they can be radiation calibration, atmospheric correction, geometric correction, noise removal, texture analysis, edge detection, deep learning classification, difference graph method, ratio method, fusion technology, etc. By publishing these processing algorithms as operator nodes / operators available in the workflow, these operator nodes can be conveniently used in the subsequent creation of custom workflows, thereby improving the creation efficiency of custom workflows and reducing the difficulty of creation.

[0020] In this embodiment, operators represent various tensor calculation operations, such as addition, multiplication, matrix multiplication, matrix transposition, etc. Operator nodes usually contain attribute information of operators and are connected to tensors involved in the operation. These tensor information are stored in the input and output of the operator nodes. The attributes of the operator include type, parameters, input and output, data type, precision, etc.

[0021] Step 12: Use the published operator nodes to create a custom workflow and save it.

[0022] In this embodiment, the operator nodes that have been published in the above steps can be used to create a custom workflow, and then the custom workflow can be saved. When creating a custom workflow, any number of the published operator nodes can be selected and reasonably combined according to different processing requirements, so that the created custom workflow has the corresponding remote sensing image data processing capabilities or methods, automatically processes remote sensing image data and generates the desired data results.

[0023] Step 13: Using the custom workflow, create a production line for automatically processing remote sensing image data, and provide the operator nodes used in the production line with data required for algorithm operation.

[0024] After creating a custom workflow, you can use it to create a production line for automated processing of remote sensing image data. Since the custom workflow contains operator nodes, the production line created using the custom workflow also uses related operator nodes. It is also necessary to provide the operator nodes used in the production line with the data required for algorithm runtime to ensure that the operator nodes can complete the relevant algorithm calculations.

[0025] Step 14: Build a process engine to provide automated operation capability for the production line, and execute each operator node in sequence according to the order of the operator nodes configured in the production line to automatically process the remote sensing image data to obtain a processing result.

[0026] In this embodiment, the process engine is constructed to provide the production line with automated operation capabilities, so that the production line executes each operator in sequence according to the operator sequence in the custom workflow, and finally obtains the processing result. Among them, the process engine is a set of code logic for system operation. It parses the data structure of the custom workflow to obtain the operation process of the production line, and controls the operation sequence and operation timing of the node class according to the process sequence, thereby realizing the automated processing of remote sensing image data.

[0027] The workflow-based automated processing method for remote sensing image data provided by the embodiment of the present invention publishes the processing algorithm of remote sensing image data as operator nodes, builds custom workflows according to different needs, and realizes automated processing of remote sensing image data, which significantly improves the processing efficiency of remote sensing image data, reduces the difficulty of operation, and has wide applicability.

[0028] The following example illustrates the above workflow-based automated processing method for remote sensing image data.

[0029] In one optional specific implementation manner, the step of publishing the remote sensing image data processing algorithm as an operator node available in the workflow includes: Upload the image used by the processing algorithm to the image warehouse, and enter the image address and image description information into the publishing parameters; Entering relevant information of the processing algorithm into the publishing parameters, the relevant information including the environment required for the processing algorithm to run, the open port of the algorithm service, the access address of the algorithm service, the storage path of the algorithm code, and the transmission parameter information of the algorithm; According to the publishing parameters, the processing algorithm is published as an operator node available in the workflow.

[0030] In this embodiment, the above steps include image information entry, algorithm information entry, and publishing as an operator. Specifically, first, the image used by the remote sensing image data processing algorithm needs to be uploaded to the image warehouse, and the image address, image description information, etc. need to be entered into the publishing parameters; then, the relevant information of the algorithm is entered into the publishing parameters, including the environment required for the algorithm to run (memory usage, CPU usage, whether to use GPU, etc.), the open port of the algorithm service, the access address of the algorithm service, the storage path of the algorithm code, the specific parameter transmission information of the algorithm, etc. Then, based on the published parameters that have been entered, the algorithm is published as an operator. When publishing, the image information and algorithm information entered in the above steps will be bound to form an operator available in the workflow.

[0031] The image description information is used to describe the image's source, purpose, and some customized remarks, so that users can easily understand what the image is used for.

[0032] In some embodiments, the step of creating and saving a custom workflow using the published operator nodes includes: Define a target data structure for storing a custom workflow, wherein the target data structure includes description information of the custom workflow, operator nodes included in the custom workflow, connection relationships between operator nodes, and data transfer relationships between operator nodes having connection relationships; Select a target operator node from the published operator nodes, and create a custom workflow according to the target operator node; Serialize the created custom workflow into the target data structure.

[0033] In this embodiment, the existing operator nodes are used to create a custom workflow, and a reasonable data structure is established to store the custom workflow. Specifically, the target data structure for storing the workflow is first defined, and the target data structure records the description information of the workflow, the operator nodes included in the workflow, the connection relationship between the operator nodes, and the data transmission between the operator nodes with the connection relationship; then, the required target operator node is selected from the published operator nodes, and then the custom workflow is created according to the actual needs. The creation process is the process of selecting the target operator node, building the connection relationship between the nodes, etc.; after the creation is completed, the custom workflow can be serialized into the data structure defined in the above steps.

[0034] Among them, serialization refers to converting structured data (such as objects, arrays, records, etc.) into a format that can be easily stored, transmitted or further processed.

[0035] In some other embodiments of the present application, the step of providing the operator nodes used in the production line with data required for algorithm operation includes: Calculating the remaining parameters of each operator node used in the production line, where the remaining parameters are other input parameters required by the operator node in addition to the data transmitted by the previous operator node; Complete the remaining parameters.

[0036] Since each operator node in the production line / workflow, in addition to using the data passed by the previous operator node, usually also requires some other input parameters to complete the algorithm that the current operator node needs to execute and then output the result, it is also necessary to calculate the remaining parameters of each operator node used in the production line. These remaining parameters are the other input parameters required by the current operator node in addition to the data passed by the previous operator node. When creating a production line, these remaining parameters need to be completed.

[0037] In some embodiments of the present application, the step of building a process engine to provide the production line with automated operation capabilities includes: Create a production line startup node class, the startup node class is used to control the startup of the production line; Creating an algorithm node class, wherein the algorithm node class is used to dynamically create the operator nodes required by the production line, and to create an algorithm container according to the publishing information of the operator nodes and complete the algorithm running process; Creating an end node class, wherein the end node class is used to control the end of the production line; A process control class is created, where the process control class is used to control the execution order of each operator node in the production line according to the process relationship of the production line.

[0038] In this embodiment, it is necessary to provide the production line with automated operation capabilities so that the production line can execute the results of each operator in sequence according to the operator sequence configured in the workflow and finally obtain the processing results. The automated operation of the production line is achieved by building a process engine. The specific construction process is as follows: 1) Create a production line startup node class to control the startup of the production line. Some pre-processing work before the production line is started is also implemented in this node class; 2) Create an algorithm node class, which is an abstract class that records the operator ID and the operator's publishing information. When the production line is running, the required operator nodes are dynamically created according to the operator ID, and the algorithm container is created according to the operator's publishing information. The algorithm running process is completed using the configured data required for the algorithm to run.

[0039] 3) Create an end node class to control the end of the production line. When the production line runs to this node, it means that there are no more nodes that need to be run. Some post-processing work of the production line is also implemented in this node class; 4) Create a process control class, which is used to ensure that each operator in the production line runs and collaborates according to business rules. When the production line starts running, this class will call the start node class. When the start node class is finished running, it will analyze the process relationship of the production line and continue to execute the downstream node pointed to by the start node. Repeat this process until it runs to the end node.

[0040] In this embodiment, the node class is usually used to define nodes in a data structure. A node is a basic unit of a data structure, which contains a data part and references (pointers or links) to other nodes. A node class usually contains the following attributes: 1. Data part: stores the value or data of the node.

[0041] 2. References / pointers: Links to other nodes, used to build data structures (such as linked lists, trees, etc.).

[0042] In this embodiment, the algorithm container is a container that actually runs the algorithm, and it can access the data path required for the algorithm to run.

[0043] See also Figure 2 , Figure 2 2 is a schematic diagram of a workflow-based automatic processing device for remote sensing image data provided by Embodiment 2 of the present invention. The device 20 includes: A publishing module 21, used to publish the processing algorithm of remote sensing image data as an operator node available in the workflow; The workflow creation module 22 is used to create and save a custom workflow using the published operator nodes; A production line creation module 23 is used to create a production line for automatically processing remote sensing image data using the custom workflow, and provide data required for algorithm operation for operator nodes used in the production line; The automated processing module 24 is used to construct a process engine to provide automated operation capabilities for the production line, and to execute each operator node in sequence according to the order of the operator nodes configured in the production line to automatically process the remote sensing image data to obtain a processing result.

[0044] Optionally, the publishing module includes: A first input unit is used to upload the image used by the processing algorithm to the image warehouse, and input the image address and image description information into the publishing parameters; A second input unit is used to input relevant information of the processing algorithm into the publishing parameters, wherein the relevant information includes the environment required for the operation of the processing algorithm, the open port of the algorithm service, the access address of the algorithm service, the storage path of the algorithm code, and the transmission parameter information of the algorithm; A publishing unit is used to publish the processing algorithm as an operator node available in the workflow according to the publishing parameters.

[0045] Optionally, the workflow creation module includes: A data structure definition unit, used to define a target data structure for storing a custom workflow, wherein the target data structure includes description information of the custom workflow, operator nodes included in the custom workflow, connection relationships between operator nodes, and data transfer relationships between operator nodes having connection relationships; A workflow creation unit, used to select a target operator node from the published operator nodes, and create a custom workflow according to the target operator node; A serialization unit is used to serialize the created custom workflow into the target data structure.

[0046] Optionally, the production line creation module includes: A calculation unit, used to calculate the remaining parameters of each operator node used in the production line, wherein the remaining parameters are other input parameters required by the operator node in addition to the data transmitted by the previous operator node; The completion unit is used to complete the remaining parameters.

[0047] Optionally, the production line creation module includes: A startup node class creation unit, used to create a production line startup node class, wherein the startup node class is used to control the startup of the production line; An algorithm node class creation unit, used to create an algorithm node class, the algorithm node class is used to dynamically create the operator nodes required by the production line, and is used to create an algorithm container according to the release information of the operator node and complete the algorithm running process; An end node class creation unit, used to create an end node class, wherein the end node class is used to control the end of the production line; The process control class creation unit is used to create a process control class, and the process control class is used to control the execution order of each operator node in the production line according to the process relationship of the production line.

[0048] In the embodiment of the present invention, by publishing the processing algorithm of remote sensing image data as an operator node and building a custom workflow according to different needs, the automatic processing of remote sensing image data is realized, which significantly improves the processing efficiency of remote sensing image data, reduces the difficulty of operation, and has wide applicability.

[0049] The embodiment of the present invention is a product embodiment corresponding to the above-mentioned method embodiment 1, so it will not be described in detail here. Please refer to the above-mentioned embodiment 1 for details.

[0050] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A workflow-based automated processing method for remote sensing image data, characterized in that: The method comprises the following steps: Publish the processing algorithms for remote sensing image data as operator nodes available in the workflow; Use the published operator nodes to create and save custom workflows; Using the custom workflow, a production line for automatically processing remote sensing image data is created, and the operator nodes used in the production line are provided with data required for algorithm operation; A process engine is constructed to provide automated operation capability for the production line, and each operator node is executed in sequence according to the order of the operator nodes configured in the production line to automatically process the remote sensing image data to obtain a processing result.

2. The method according to claim 1, characterized in that: The step of publishing the remote sensing image data processing algorithm as an operator node available in the workflow includes: Upload the image used by the processing algorithm to the image warehouse, and enter the image address and image description information into the publishing parameters; Entering relevant information of the processing algorithm into the publishing parameters, the relevant information including the environment required for the processing algorithm to run, the open port of the algorithm service, the access address of the algorithm service, the storage path of the algorithm code, and the transmission parameter information of the algorithm; According to the publishing parameters, the processing algorithm is published as an operator node available in the workflow.

3. The method according to claim 1, characterized in that The steps of creating and saving a custom workflow using the published operator nodes include: Define a target data structure for storing a custom workflow, wherein the target data structure includes description information of the custom workflow, operator nodes included in the custom workflow, connection relationships between operator nodes, and data transfer relationships between operator nodes having connection relationships; Select a target operator node from the published operator nodes, and create a custom workflow according to the target operator node; Serialize the created custom workflow into the target data structure.

4. The method according to claim 1, characterized in that The step of providing the operator nodes used in the production line with data required for algorithm operation includes: Calculating the remaining parameters of each operator node used in the production line, where the remaining parameters are other input parameters required by the operator node in addition to the data transmitted by the previous operator node; Complete the remaining parameters.

5. The method according to claim 1, characterized in that The step of building a process engine to provide the production line with automated operation capabilities includes: Create a production line startup node class, the startup node class is used to control the startup of the production line; Creating an algorithm node class, wherein the algorithm node class is used to dynamically create the operator nodes required by the production line, and to create an algorithm container according to the publishing information of the operator nodes and complete the algorithm running process; Creating an end node class, wherein the end node class is used to control the end of the production line; A process control class is created, where the process control class is used to control the execution order of each operator node in the production line according to the process relationship of the production line.

6. A workflow-based remote sensing image data automatic processing device, characterized in that: The device comprises: The publishing module is used to publish the processing algorithms of remote sensing image data as operator nodes available in the workflow; The workflow creation module is used to create and save custom workflows using published operator nodes; A production line creation module, used to create a production line for automatically processing remote sensing image data using the custom workflow, and provide the operator nodes used in the production line with data required for algorithm operation; The automated processing module is used to construct a process engine to provide automated operation capabilities for the production line, and to execute each operator node in sequence according to the order of the operator nodes configured in the production line to automatically process the remote sensing image data to obtain a processing result.

7. The device according to claim 6, characterized in that The publishing module includes: A first input unit is used to upload the image used by the processing algorithm to the image warehouse, and input the image address and image description information into the publishing parameters; A second input unit is used to input relevant information of the processing algorithm into the publishing parameters, wherein the relevant information includes the environment required for the operation of the processing algorithm, the open port of the algorithm service, the access address of the algorithm service, the storage path of the algorithm code, and the transmission parameter information of the algorithm; A publishing unit is used to publish the processing algorithm as an operator node available in the workflow according to the publishing parameters.

8. The device according to claim 6, characterized in that The workflow creation module includes: A data structure definition unit, used to define a target data structure for storing a custom workflow, wherein the target data structure includes description information of the custom workflow, operator nodes included in the custom workflow, connection relationships between operator nodes, and data transfer relationships between operator nodes having connection relationships; A workflow creation unit, used to select a target operator node from the published operator nodes, and create a custom workflow according to the target operator node; A serialization unit is used to serialize the created custom workflow into the target data structure.

9. The device according to claim 6, characterized in that The production line creation module includes: A calculation unit, used to calculate the remaining parameters of each operator node used in the production line, wherein the remaining parameters are other input parameters required by the operator node in addition to the data transmitted by the previous operator node; The completion unit is used to complete the remaining parameters.

10. The device according to claim 6, characterized in that The production line creation module includes: A startup node class creation unit, used to create a production line startup node class, wherein the startup node class is used to control the startup of the production line; An algorithm node class creation unit, used to create an algorithm node class, the algorithm node class is used to dynamically create the operator nodes required by the production line, and is used to create an algorithm container according to the release information of the operator node and complete the algorithm running process; An end node class creation unit, used to create an end node class, wherein the end node class is used to control the end of the production line; The process control class creation unit is used to create a process control class, and the process control class is used to control the execution order of each operator node in the production line according to the process relationship of the production line.

Images