A flink-based real-time radar data three-dimensional point cloud presentation method and system

Abstract

The application provides a kind of based on Flink's real-time radar data three-dimensional point cloud presentation method and system, the method includes: obtaining monitoring data transmission to distributed message queue kafka, and obtaining radar data conversion to Potree required format transmission to front end;Based on Flink, the data in distributed message queue kafka is handled and associated with table, and stored in database, while according to the current monitoring data transmission to PyTorch prediction model of the region of interest according to business needs;Through the prediction model, according to the rules or pre-trained model preset, the current monitoring data is processed, and the early warning information is output to the front end;According to the business requirement, the historical data in the database is inquired to carry out real-time analysis, and the data report is generated and sent to the front end;Front end is presented to radar data by Potree tool, and the prediction conclusion and data report are displayed.The application supports high throughput, low delay, high performance transmission of real-time point cloud data, improves data access efficiency and speed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radar data processing, and in particular to a three-dimensional point cloud presentation method and system for real-time radar data based on Flink. BACKGROUND

[0002] Point cloud data refers to a set of vectors in a three-dimensional coordinate system, and in addition to geometric positions, some point cloud data also has color information. Laser radar can achieve dense sampling of the earth's surface and generate high-precision three-dimensional x, y, z measurement values, and can generate discrete multi-point cloud data sets that can be managed, displayed, analyzed and shared through GIS.

[0003] In geological disasters, the application of radar technology is becoming more and more common, and higher requirements are put forward for real-time rendering of radar point cloud data. It is hoped that the point cloud data collected by the radar device can be quickly viewed on the web, which is particularly important for the prevention and control of geological disasters.

[0004] The commonly used web point cloud loading method at present is cesium, but cesium has poor performance when the data volume is large. Therefore, some use Potree to realize data viewing. Potree is a very large dataset point cloud viewer based on WebGL. Potree software can render large laser radar point clouds on the web, but the problem of high throughput and high performance transmission of point cloud data has not been solved, and the online real-time rendering effect required by customers cannot be achieved. SUMMARY

[0005] The technical problem to be solved by the present application is to provide a three-dimensional point cloud presentation method and system for real-time radar data based on Flink, to improve data access efficiency and speed, realize efficient presentation of three-dimensional point clouds, and improve user experience.

[0006] In a first aspect, the present application provides a three-dimensional point cloud presentation method for real-time radar data based on Flink, comprising the following steps:

[0007] Step S1, acquiring monitoring data, transmitting all monitoring data to a distributed message queue kafka, and acquiring radar data therein and converting it into a format required by Potree and passing it to the front end;

[0008] Step S2, based on Flink, performing stream batch integrated real-time calculation processing on the data in the distributed message queue kafka and associating with the dimension table, storing the processed data in the database, and simultaneously acquiring the current monitoring data of the region of interest according to the business needs and transmitting it to the PyTorch prediction model;

[0009] Step S3, the PyTorch prediction model processes the current monitoring data according to the pre-set rules or pre-trained model, and outputs the prediction conclusion to the front end;

[0010] Step S4, according to the business requirement, the historical data is queried from the database for real-time analysis, and the data report is generated and sent to the front end;

[0011] Step S5, the front end presents the radar data in three dimensions through the Potree tool, and displays the prediction conclusion and data report of the region of interest.

[0012] Further, the step S1 of "obtaining radar data and converting it into a format required by Potree and passing it to the front end" specifically includes: after converting the radar data into a format required by Potree, storing it in a pre-built minio distributed server, and calling the corresponding data in the minio distributed server when the front end needs to present real-time three-dimensional point cloud.

[0013] Further, the step S2 specifically includes:

[0014] The radar data is read from the distributed message queue kafka through the Flink cluster, the data is decomposed through the operator of the Flink cluster, stored in the Elasticsearch database, and the attribute data is stored in the PostgreSQL database;

[0015] The monitoring data obtained from different devices is associated and stored in the Elasticsearch database and the PostgreSQL database;

[0016] According to the business requirement, the current monitoring data of the region of interest is obtained and transmitted to the PyTorch prediction model.

[0017] Further, the step 3 specifically includes:

[0018] If the monitoring data only includes radar data, the PyTorch prediction model obtains the current data of the region of interest according to the business requirement, compares the data deformation amount of the current data and the historical data with the pre-set threshold value, and outputs the warning level of the current region to the front end;

[0019] If the monitoring data includes other associated monitoring data in addition to radar data, the PyTorch prediction model is pre-trained according to each monitoring data to form a mountain collapse prediction model, a landslide prediction model, a debris flow prediction model, a ground subsidence prediction model, a ground fissure prediction model or a ground subsidence prediction model required by the business, and is stored, and when the user selects the region of interest, the corresponding prediction model is called to output the corresponding warning level to the front end;

[0020] The monitoring data includes data collected by one or more devices of a rain gauge, a crack meter, a ground sound meter, a ground electricity meter and a vibration meter according to a user self-defined configuration in addition to radar data monitored by radar.

[0021] Further, the data report in the step S4 is a real-time change graph of monitoring data obtained by other devices in addition to radar data, including a curve graph, a column graph or a pie graph.

[0022] In a second aspect, the present application provides a three-dimensional point cloud presentation system of real-time radar data based on Flink, which comprises a data acquisition module, a batch processing module, a prediction module, an analysis module and a three-dimensional point cloud presentation module.

[0023] The data acquisition module is configured to acquire monitoring data, transmit all monitoring data to a distributed message queue Kafka, and convert radar data in the distributed message queue Kafka into a format required by Potree and transmit the radar data to a front end.

[0024] The batch processing module is configured to perform real-time computing and table association of data in the distributed message queue Kafka based on Flink, store processed data in a database, and transmit current monitoring data of a region of interest to a PyTorch prediction model according to business needs.

[0025] The prediction module is configured to process current monitoring data according to a pre-set rule or a pre-trained model by the PyTorch prediction model, and output a prediction conclusion to the front end.

[0026] The analysis module is configured to query historical data from the database for real-time analysis, and generate a data report and transmit the data report to the front end according to business needs.

[0027] The three-dimensional point cloud presentation module is configured to perform three-dimensional presentation of radar data by the front end through a Potree tool, and display a prediction conclusion of a region of interest and a data report.

[0028] Further, the data acquisition module comprises the following steps: converting radar data into a format required by Potree, storing the radar data in a pre-built minio distributed server, and calling corresponding data in the minio distributed server when real-time three-dimensional point cloud presentation is required in the front end.

[0029] Further, the batch processing module comprises the following steps:

[0030] Radar data is read from a distributed message queue kafka by a Flink cluster, the data is decomposed by an operator of the Flink cluster, stored in an Elasticsearch database, and attribute data in the database is stored in a PostgreSQL database;

[0031] The monitoring data obtained from different devices are associated and stored in the Elasticsearch database and the PostgreSQL database;

[0032] According to the business needs, the current monitoring data of the region of interest are transmitted to the PyTorch prediction model.

[0033] Further, the prediction module specifically comprises:

[0034] If the monitoring data only includes radar data, the PyTorch prediction model obtains the current data of the region of interest according to the business needs, compares the data deformation amount of the current data and historical data with a preset threshold value, and outputs the early warning level of the current region to the front end;

[0035] If the monitoring data includes other associated monitoring data in addition to radar data, the PyTorch prediction model is pre-stored according to the training of each monitoring data into a mountain collapse prediction model, a landslide prediction model, a debris flow prediction model, a ground subsidence prediction model, a ground fissure prediction model or a ground subsidence prediction model required by the business, and when the user selects the region of interest, the corresponding prediction model is called to output the corresponding early warning level to the front end;

[0036] In addition to the radar data monitored by the radar, the monitoring data also includes data collected by one or more devices of a rain gauge, a crack meter, a ground sound meter, a ground electricity meter and a vibration meter according to user self-defined configuration.

[0037] Further, the data report in the analysis module is a real-time change graph of the monitoring data obtained by the other devices in addition to the radar data, including a curve graph, a column graph or a pie chart.

[0038] One or more technical solutions provided by the present application have at least the following technical effects or advantages: based on Flink, a large amount of data is processed in real time online, the high-throughput, low-latency and high-performance transmission requirements of real-time point cloud data are met, distributed storage and query data are realized, real-time analysis and early warning output according to business needs are realized, the front end uses Potree to quickly display and operate the three-dimensional point cloud model on the Web, and the analysis and early warning data required by the user are synchronously and online displayed in real time, and the user experience is improved. BRIEF DESCRIPTION OF DRAWINGS

[0039] The application will be further described below with reference to the accompanying drawings and embodiments.

[0040] Figure 1 An execution flow chart of the three-dimensional point cloud rendering method based on Flink real-time radar data of the application.

[0041] Figure 2 An execution flow chart of the three-dimensional point cloud rendering method based on Flink real-time radar data of the preferred embodiment of the application.

[0042] Figure 3 A structural schematic diagram of the three-dimensional point cloud rendering system based on Flink real-time radar data of the application. DETAILED DESCRIPTION

[0043] The embodiments of the application provide a three-dimensional point cloud rendering method and system based on Flink real-time radar data, which are used to solve the problems of low transmission performance and poor user online real-time viewing experience in the existing real-time radar data three-dimensional rendering.

[0044] In order to better understand the above technical solutions, the above technical solutions will be described in detail below in combination with the accompanying drawings and specific embodiments of the specification.

[0045] Embodiment one

[0046] The embodiment provides a three-dimensional point cloud rendering method based on Flink real-time radar data, which comprises the following steps as shown in Figure 1 and Figure 2 .

[0047] Step S1, acquiring monitoring data, transmitting all monitoring data to a distributed message queue kafka, and acquiring radar data therein and converting the radar data into a format required by Potree and delivering the radar data to a front end;

[0048] Step S2, based on Flink, performing stream batch integrated real-time calculation processing on data in the distributed message queue kafka and associating with a dimension table, storing the processed data in a database, and simultaneously acquiring current monitoring data of a region of interest according to business needs and transmitting the current monitoring data to a PyTorch prediction model;

[0049] Step S3, processing the current monitoring data according to a pre-set rule or a pre-trained model by the PyTorch prediction model, and outputting a prediction conclusion to the front end;

[0050] Step S4, querying historical data from the database according to business needs for real-time analysis, and generating a data report and sending the data report to the front end;

[0051] Step S5, the front end presents the radar data in three dimensions through the Potree tool, and displays the prediction conclusion and data report of the region of interest.

[0052] Preferably, the step S1 of "obtaining radar data and converting the radar data into a format required by Potree and transmitting the radar data to the front end" specifically comprises: storing the radar data converted into a format required by Potree into a pre-built minio distributed server, and calling the corresponding data in the minio distributed server when the front end needs to present real-time three-dimensional point cloud.

[0053] Preferably, the step S2 specifically comprises:

[0054] The radar data is read from the distributed message queue kafka through the Flink cluster, the data is decomposed through the operator of the Flink cluster, stored in the Elasticsearch database, and the attribute data is stored in the PostgreSQL database;

[0055] The monitoring data obtained from different devices is associated and stored in the Elasticsearch database and the PostgreSQL database;

[0056] According to the business needs, the current monitoring data of the region of interest is obtained and transmitted to the PyTorch prediction model.

[0057] Preferably, the step 3 specifically comprises:

[0058] If the monitoring data only includes radar data, the PyTorch prediction model obtains the current data of the region of interest according to the business needs, compares the data deformation amount of the current data and historical data with a preset threshold value, and outputs the warning level of the current region to the front end;

[0059] If the monitoring data includes other associated monitoring data (such as rainfall data monitored by a rain gauge, crack data monitored by a crack meter, etc.) in addition to radar data, the PyTorch prediction model is pre-stored according to each monitoring data to form a mountain collapse prediction model, a landslide prediction model, a debris flow prediction model, a ground subsidence prediction model, a ground crack prediction model or a ground subsidence prediction model required by the business, and outputs the corresponding warning level to the front end through the calling of the trained corresponding prediction model when the user selects the region of interest.

[0060] Preferably, the monitoring data includes data collected by one or more of rain gauges, crack meters, ground sound meters, ground electricity meters, vibration meters and other monitoring instruments configured by the user.

[0061] Preferably, the data report in the step S4 is a real-time change graph of the monitoring data obtained by the devices other than the radar data, including a curve graph, a column graph or a pie graph, which can provide an analysis basis for the user to review and warn information.

[0062] Based on the same inventive concept, the application also provides a system corresponding to the method in embodiment one, which is described in detail in embodiment two.

[0063] Embodiment two

[0064] In this embodiment, a three-dimensional point cloud presentation system for real-time radar data based on Flink is provided, as shown in Figure 3 The system includes a data acquisition module, a batch processing module, a prediction module, an analysis module and a three-dimensional point cloud presentation module.

[0065] The data acquisition module is configured to acquire monitoring data, transmit all monitoring data to a distributed message queue kafka, and convert radar data in the monitoring data into a format required by Potree and transmit the radar data to a front end.

[0066] The batch processing module is configured to perform real-time computing and processing on data in the distributed message queue kafka based on Flink, associate the data with a dimension table, store the processed data in a database, and transmit current monitoring data of a region of interest to a PyTorch prediction model according to business needs.

[0067] The prediction module is configured to process the current monitoring data according to a pre-set rule or a pre-trained model by using the PyTorch prediction model, and output a prediction conclusion to the front end.

[0068] The analysis module is configured to query historical data from the database for real-time analysis according to business needs, and generate a data report and transmit the data report to the front end.

[0069] The three-dimensional point cloud presentation module is configured to perform three-dimensional presentation on the radar data by using a Potree tool in the front end, and display the prediction conclusion of the region of interest and the data report.

[0070] Preferably, the data acquisition module includes the following steps: converting the radar data into the format required by Potree, storing the converted data in a pre-built minio distributed server, and calling the corresponding data in the minio distributed server when the front end needs to present real-time three-dimensional point cloud.

[0071] Preferably, the batch processing module includes the following steps:

[0072] The Flink cluster reads the radar data from the distributed message queue kafka, decomposes the data through the operator of the Flink cluster, stores the data in the Elasticsearch database, and stores the attribute data in the PostgreSQL database.

[0073] The monitoring data obtained from different devices are associated and stored in the Elasticsearch database and the PostgreSQL database.

[0074] According to the business needs, the current monitoring data of the region of interest are transmitted to the PyTorch prediction model.

[0075] Preferably, the prediction module includes the following steps:

[0076] If the monitoring data only includes radar data, the PyTorch prediction model compares the data deformation amount of the current data and historical data with a preset threshold according to the business needs, and outputs the warning level of the current region to the front end.

[0077] If the monitoring data includes other associated monitoring data (such as rainfall data monitored by a rain gauge, crack data monitored by a crack meter, etc.) in addition to radar data, the PyTorch prediction model is pre-stored according to the business needs of a mountain collapse prediction model, a landslide prediction model, a debris flow prediction model, a ground subsidence prediction model, a ground crack prediction model, or a ground subsidence prediction model, and outputs the corresponding warning level to the front end by calling the trained corresponding prediction model when the user selects the region of interest.

[0078] Preferably, the monitoring data includes data collected by one or more of the monitoring instruments such as a rain gauge, a crack meter, a ground sound meter, a ground electric meter, and a vibration meter, in addition to the radar data monitored by a radar. According to user needs, when the user connects multiple monitoring instruments, the PyTorch prediction model is dynamically adjusted according to the data obtained by the connected instrument devices, ensuring the accuracy and pertinence of the PyTorch prediction model.

[0079] Preferably, the data report in the analysis module is a real-time change graph of monitoring data obtained by other devices except radar data, including a curve graph, a column graph or a pie chart, which can provide analysis basis for the user to review and warn information.

[0080] Since the system introduced in the second embodiment of the present application is the system used for implementing the method of the first embodiment of the present application, the specific structure and variations of the system can be understood by those skilled in the art based on the method introduced in the first embodiment of the present application, and thus will not be described here again. Any system used for the method of the first embodiment of the present application belongs to the scope of the present application.

[0081] The present application realizes real-time online processing of point cloud big data based on Flink, supports high-throughput, low-latency and high-performance transmission of real-time point cloud data, and realizes real-time analysis and output of data through distributed storage and query data, the front end uses Potree to quickly display and operate the three-dimensional point cloud model on the Web, and provides online data analysis and early warning functions for users, reports required for real-time prediction information and review analysis of users, and improves user experience.

[0082] Although the specific embodiments of the present application are described above, those skilled in the art should understand that the specific embodiments described are only illustrative, and are not intended to limit the scope of the present application, and equivalent modifications and changes made by those skilled in the art in accordance with the spirit of the present application should be covered within the scope of the claims of the present application.

Claims

1. A method for rendering a three-dimensional point cloud of real-time radar data based on Flink, characterized in that: It comprises the following steps: Step S1, obtaining monitoring data, transmitting all monitoring data to a distributed message queue Kafka, and obtaining radar data therein and converting the radar data into a format required by Potree and delivering the radar data to a front end; Step S2, based on Flink, performing stream batch integrated real-time calculation processing on data in the distributed message queue Kafka and associating with a dimension table, storing the processed data in a database, and simultaneously obtaining current monitoring data of a region of interest according to business needs and delivering the monitoring data to a PyTorch prediction model; Step S3, processing the current monitoring data according to a pre-set rule or a pre-trained model by the PyTorch prediction model, and outputting a prediction conclusion to the front end; Step S4, querying historical data from the database according to business needs for real-time analysis, and generating a data report and sending the data report to the front end; Step S5, the front end performs three-dimensional presentation on the radar data by using a Potree tool, and displays the prediction conclusion of the region of interest and the data report.

2. The method of claim 1, wherein: The "obtaining radar data therein and converting the radar data into a format required by Potree and delivering the radar data to a front end" in the step S1 specifically comprises: after converting the radar data into a format required by Potree, storing the radar data in a pre-built minio distributed server, and when real-time three-dimensional point cloud needs to be presented in the front end, calling corresponding data in the minio distributed server.

3. The method of claim 1, wherein: The step S2 specifically comprises: reading radar data from the distributed message queue Kafka through a Flink cluster, decomposing the data through an operator of the Flink cluster, storing the data in an Elasticsearch database, and storing attribute data therein in a PostgreSQL database; associating monitoring data obtained from different devices and storing the monitoring data in the Elasticsearch database and the PostgreSQL database; according to business needs, obtaining current monitoring data of a region of interest and delivering the monitoring data to a PyTorch prediction model.

4. The method of claim 1, wherein: The step S3 specifically comprises: if the monitoring data only includes radar data, the PyTorch prediction model obtains current data of a region of interest according to business needs, compares the data deformation amount of the current data and historical data with a pre-set threshold value, and outputs a warning level of the current region to the front end; if the monitoring data includes other associated monitoring data in addition to the radar data, the PyTorch prediction model pre-stores a mountain collapse prediction model, a landslide prediction model, a debris flow prediction model, a ground subsidence prediction model, a ground fissure prediction model, or a ground subsidence prediction model trained according to each monitoring data according to business needs, and when a user selects a region of interest, outputs a corresponding warning level to the front end by calling the trained corresponding prediction model; in addition to the radar data monitored by the radar, the monitoring data also includes data collected by one or more devices of a rain gauge, a crack meter, a ground sound meter, a ground electricity meter, and a vibration meter according to user self-defined configuration.

5. The method of claim 1, wherein: The data report in the step S4 is a real-time change graph of the monitoring data obtained by other devices except radar data, including a curve graph, a column graph or a pie graph. 6.A Flink-based real-time radar data three-dimensional point cloud rendering system, characterized in that: The system comprises a data acquisition module, a batch processing module, a prediction module, an analysis module and a three-dimensional point cloud presentation module: The data acquisition module is configured to acquire monitoring data, transmit all the monitoring data to a distributed message queue Kafka, and acquire radar data in the distributed message queue Kafka, convert the radar data into a format required by Potree, and transmit the radar data to a front end. The batch processing module is configured to perform real-time calculation and processing on data in the distributed message queue Kafka based on Flink, associate the data with a dimension table, store the processed data in a database, and acquire current monitoring data of a region of interest according to a business requirement and transmit the current monitoring data to a PyTorch prediction model. The prediction module is configured to process the current monitoring data according to a preset rule or a pre-trained model by using the PyTorch prediction model, and output a prediction conclusion to the front end. The analysis module is configured to query historical data from the database according to a business requirement, perform real-time analysis, and generate a data report and transmit the data report to the front end. The three-dimensional point cloud presentation module is configured to perform three-dimensional presentation on the radar data by using a Potree tool on the front end, and display the prediction conclusion of the region of interest and the data report.

7. The system of claim 6, wherein: The data acquisition module specifically comprises the following steps: storing radar data in a format required by Potree in a minio distributed server in advance after converting the radar data into the format required by Potree, and calling corresponding data in the minio distributed server when real-time three-dimensional point cloud presentation is required on the front end.

8. The system of claim 6, wherein: The batch processing module specifically comprises the following steps: reading radar data from the distributed message queue Kafka by using a Flink cluster, decomposing the data by using an operator of the Flink cluster, storing the data in an Elasticsearch database, and storing attribute data in the radar data in a PostgreSQL database; associating monitoring data obtained from different devices and storing the monitoring data in the Elasticsearch database and the PostgreSQL database; acquiring current monitoring data of a region of interest according to a business requirement and transmitting the current monitoring data to a PyTorch prediction model.

9. The system of claim 6, wherein: The prediction module specifically comprises the following steps: if the monitoring data only comprises radar data, the PyTorch prediction model acquires current data of a region of interest according to a business requirement, compares a data deformation amount of the current data and historical data with a preset threshold value, and outputs a warning level of the current region to the front end; if the monitoring data comprises other associated monitoring data except radar data, the PyTorch prediction model stores a mountain collapse prediction model, a landslide prediction model, a debris flow prediction model, a ground subsidence prediction model, a ground fissure prediction model or a ground subsidence prediction model trained according to the monitoring data, and outputs a corresponding warning level to the front end by calling the corresponding prediction model when a user selects a region of interest. The monitoring data includes data collected by one or more devices configured by the user, in addition to radar data monitored by the radar.

10. The system of claim 6, wherein: The data report in the analysis module is a real-time change graph of monitoring data obtained by devices other than radar, including a curve graph, a column graph or a pie chart.

Images