Pixel-level convection cloud labeling system based on FY-4A

Abstract

The invention provides a pixel-level convection cloud labeling system based on FY-4A, and belongs to the technical field of meteorological monitoring and data processing, and the system comprises a software interface and a data processing module. The software interface provides a visual operation platform for a user, the main image display frame is used for displaying an example that a marking result covers a brightness temperature image, and the three auxiliary image display frames sequentially display a radar echo image, a visible light image and a sequence brightness temperature color image from top to bottom; the data processing module is used for processing and analyzing the FY-4A satellite data, and the data reading sub-module is used for reading the FY-4A satellite data; the threshold calculation sub-module calculates an annotation threshold of the streaming cloud according to the parameters set by the user and the satellite data; the labeling result generation sub-module labels the satellite data according to the calculated threshold value, and feeds back a labeling result to the software interface for display; according to the method, the user can be helped to more accurately judge the position and the range of the convection cloud, and the accuracy and the efficiency of labeling are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of meteorological monitoring and data processing, and specifically relates to a pixel-level convective cloud labeling system based on FY-4A. BACKGROUND

[0002] In meteorological research, the formation, development and evolution of convective clouds are important research topics. As the carrier of severe convective weather, the internal physical process of convective clouds is complex and has an important influence on the development and change of weather systems. The pixel-level convective cloud labeling tool based on FY-4A of the present application can provide meteorological researchers with high-precision convective cloud data labeling. Through analysis of these labeled data, researchers can gain an in-depth understanding of the microstructure and macro features of convective clouds, such as the cloud top height, cloud bottom temperature, and water vapor content within the cloud. These information helps to reveal the formation mechanism of convective clouds, such as how convective clouds are triggered and developed under different atmospheric circulation backgrounds, and the interaction between convective clouds and the surrounding environment. In addition, by analyzing the labeled data over a long time series, the seasonal and interannual variation rules of convective clouds can also be studied, providing important basic data for climate change research.

[0003] Severe convective weather such as heavy rain, lightning, gale, and hail often poses a serious threat to people's life and property safety. Accurate and timely meteorological disaster warning is the key to reducing disaster losses. The labeling tool can help meteorological departments more accurately detect the location, range and development trend of severe convective clouds. In the process of meteorological disaster warning, through real-time labeling and analysis of FY-4A satellite data, potential severe convective weather systems can be detected in time. For example, when the labeling tool detects features such as a sharp drop in convective cloud top temperature and rapid development of cloud body, it indicates that severe convective weather may occur soon. Meteorological departments can issue warning signals in advance based on this information to remind the public to take precautions. At the same time, the labeling tool can also provide accurate input data for meteorological disaster warning models, improving the prediction accuracy of the warning models, and thus better protecting people's life and property safety.

[0004] Climate change is one of the major challenges facing the world today. As an important part of atmospheric circulation and water cycle, the activity of convective clouds is closely related to climate change. The labeling tool of the present application can provide long-term and accurate convective cloud data for climate analysis. By statistically analyzing the labeled data of convective clouds in different regions and time periods, the relationship between the distribution characteristics and change trend of convective clouds and climate change can be studied. For example, the distribution changes of convective clouds in different seasons and latitudes, and the correlation between convective cloud activity and global temperature, precipitation and other climate factors are analyzed. These research results help to deepen the understanding of the mechanism of climate change, and provide scientific basis for climate prediction and response to climate change.

[0005] Threshold-based strong convective cloud detection algorithms are commonly used convective cloud detection methods, such as the brightness temperature (BT) threshold method which uses the relatively low cloud top temperature in strong convective cloud clusters to detect strong convective cloud regions, and the threshold is usually set to 208K to 255K. However, this threshold is not universal, and the value range is affected by weather conditions, cloud physical properties, and satellite image resolution. Threshold method is essentially a rule-based algorithm that needs to set the value range of physical variables in advance, and classify each point on the satellite cloud image, and the area less than the specified threshold is determined as a strong convective cloud. When using the brightness temperature threshold method to label convective clouds, different labeling thresholds need to be set for different areas of convective clouds through manual screening, which is tedious and inefficient.

[0006] There is currently a lack of a pixel-level convective cloud labeling tool specifically for FY-4A data. FY-4A satellite has unique observation capabilities and data characteristics, but existing labeling tools cannot fully utilize its advantages and cannot meet the needs of convective cloud research and monitoring based on FY-4A data, which has a huge impact on related research. SUMMARY

[0007] To solve the above problems, the present application provides a pixel-level convective cloud labeling system based on FY-4A, which is based on the brightness temperature threshold method, uses the different spectral characteristics of different types of clouds in multispectral remote sensing images, i.e. the reflectivity and low temperature characteristics of clouds, to construct several spectral threshold conditions, and finally classifies the clouds by the corresponding threshold.

[0008] The present application is implemented by the following technical solutions: a pixel-level convective cloud labeling system based on FY-4A: The system includes a software interface and a data processing module; The software interface provides a visual operation platform for users, including a main image display box and three sub-image display boxes, the main image display box is used to show examples of labeling results overlaid on brightness temperature images, and the three sub-image display boxes display radar echo images, visible light images and sequence brightness temperature color maps from top to bottom; The data processing module is connected to the software interface through a data transmission interface, and processes and analyzes FY-4A satellite data, including a data reading submodule, a threshold calculation submodule and a labeling result generation submodule; The data reading submodule is used to read FY-4A satellite data; The threshold calculation submodule calculates the labeling threshold of the convective cloud according to the parameters set by the user and the satellite data; The labeling result generation submodule labels the satellite data according to the calculated threshold value, and feeds back the labeling result to the software interface for display.

[0009] Further, the sequence brightness temperature color chart is used to assist the labeling personnel in selecting a segmentation threshold.

[0010] Further, the software interface further comprises an operation button group, specifically: a data selection button, an image switching button, a region selection button, and a threshold adjustment button; for selecting a folder, the previous image, the next image, switching channels, a rectangle, a polygon, clearing, and a probe operation.

[0011] Further, the data labeling process of the data processing module is: Pre-segmentation: generating a red pre-labeling region according to the built-in threshold table when loading data; Delete non-convective regions: the user selects non-convective clouds or dissipating convective clouds in the pre-segmentation through the region selection button and deletes them; Fine-tune the threshold: adjust the threshold to separate multi-peak targets or remove scattered points, and retain the main body of the convective cloud.

[0012] Further, in the data labeling process, the threshold selection change amplitude of the same convective cloud target between adjacent frames is maintained between 2K-5K.

[0013] Further, the data labeling convective cloud labeling threshold is divided by region: Southern region: 215K-225K; Northern region: 220K-240K.

[0014] Further, in the data labeling process, by adjusting the brightness temperature threshold, the convective cloud groups with different brightness temperature peaks and the cirrus water vapor are distinguished by obvious boundaries, and the temperature of the convective cloud is lower than that of the cirrus or water vapor cloud group; the temperature of different convective clouds also has different differences according to the region.

[0015] A pixel-level convective cloud labeling method based on FY-4A: The method specifically comprises the following steps: Step 1: loading FY-4A AGRI data through the software interface, automatically performing pre-segmentation, and generating an initial labeling region in the main image display box; Step 2: verifying the initial labeling region based on the radar echo image, visible light image, and sequence brightness temperature color chart of the sub-image display box, and deleting non-convective clouds and dissipating convective clouds; Step 3: fine-tuning the brightness temperature threshold to separate multi-peak targets and remove scattered labeling points; Step 4: generating a pixel-level convective cloud labeling result.

[0016] An electronic device comprising a memory storing a computer program and a processor implementing the steps of the above method when executing the computer program.

[0017] A computer readable storage medium for storing computer instructions which, when executed by a processor, implement the steps of the above method.

[0018] Advantages of the present application The present application is based on the brightness temperature threshold method, which uses the different spectral characteristics of different types of clouds in multispectral remote sensing images, i.e. the reflectivity and low temperature characteristics of clouds, to construct several spectral threshold conditions, and finally classifies the clouds by corresponding thresholds. In the labeling process, through the operation of the software interface, the user can conveniently set and adjust the labeling threshold, and the data processing module processes and analyzes the satellite data according to the threshold set by the user, realizing the pixel-level labeling of convective clouds. At the same time, the radar echo image, visible light image and sequence brightness temperature color map provided by the software interface help the user to more accurately determine the position and range of convective clouds, and improve the accuracy and efficiency of labeling.

[0019] 1. Improve labeling efficiency In the prior art, when using the brightness temperature threshold method for convective cloud labeling, different labeling thresholds need to be set for different areas of convective clouds through manual screening, and the labeling process is tedious. The labeling tool of the present application provides a visual operation interface, and the user can complete the labeling of convective clouds through simple click and drag operations, greatly simplifying the labeling process and realizing efficient data labeling. The time for labeling one image is on average within two minutes, and compared with the traditional method, the labeling efficiency has been significantly improved.

[0020] 2. Improve labeling accuracy The labeling tool of the present application combines radar echo images, visible light images, sequence brightness temperature color maps and other multi-source information to help labeling personnel more accurately determine the position and range of convective clouds. When labeling convective clouds, the brightness temperature threshold is adjusted to try to distinguish convective cloud clusters with different brightness temperature peaks from cirrus water vapor, so that they have relatively obvious boundaries. At the same time, during data labeling, the continuity of the same convective cloud cluster is ensured as much as possible, and the threshold selection change amplitude of the same convective cloud target between adjacent frames is maintained between 2K~5K, improving the accuracy of labeling.

[0021] 3. Strong pertinence The labeling tool of the present application is specially developed for FY-4A data, and fully utilizes the observation ability and data characteristics of FY-4A satellite. The tool labels the pixel-level convective cloud meteorological data in China region based on the AGRI data of FY-4A, and can better meet the needs of developing convective cloud research and monitoring based on FY-4A data.

[0022] In summary, the present application can be applied to the fields of meteorological research, meteorological disaster warning, climate analysis, etc. In meteorological research, it helps to better understand the formation, development and evolution law of convective clouds; in meteorological disaster warning, it can more accurately detect severe convective weather, issue early warning and reduce disaster losses; in climate analysis, it provides data support for analyzing the influence of climate change on convective clouds. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 It is an example diagram of the overall page of the labeling system of the present application.

[0024] Figure 2 It is a schematic diagram of the front-end and back-end interaction process of the labeling system of the present application.

[0025] Figure 3 It is the initial interface of the labeling system of the present application.

[0026] Figure 4 It is an example diagram of data directory selection of the labeling system of the present application.

[0027] Figure 5 It is an example diagram of date selection of the labeling system of the present application.

[0028] Figure 6 It is an example diagram of labeling time selection of the labeling system of the present application.

[0029] Figure 7 It is an example diagram of judging pre-segmentation result of the labeling system of the present application.

[0030] Figure 8 It is an example diagram of labeling by adjusting threshold value using the labeling system of the present application.

[0031] Figure 9 It is an example diagram of removing redundant labeling results using the labeling system of the present application.

[0032] Figure 10 It is a convective cloud brightness temperature threshold segmentation diagram using the labeling system of the present application.

[0033] Figure 11 It is an example diagram of convective cloud labeling result of the labeling system of the present application, wherein (a) is a brightness temperature channel labeling result diagram, and (b) is a convective cloud label result diagram. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0035] The experimental methods used in the following embodiments are conventional methods without special instructions. The materials, reagents, methods and instruments used are conventional materials, reagents, methods and instruments in the art without special instructions, and can be obtained by a person of ordinary skill in the art through commercial channels.

[0036] The labeling tool of the present application is based on the brightness temperature threshold method, uses the characteristic performance of different categories of clouds in different bands in multispectral remote sensing images, that is, the reflectivity and low temperature characteristics of clouds, constructs a plurality of spectral threshold conditions, and finally classifies the clouds through the corresponding threshold. In the labeling process, through the operation of the software interface, the user can conveniently set and adjust the labeling threshold, and the data processing module processes and analyzes the satellite data according to the threshold set by the user, so as to realize the pixel-level labeling of the convective cloud. At the same time, the radar echo image, visible light image and sequence brightness temperature color map and other information provided by the software interface help the user to more accurately judge the position and range of the convective cloud, and improve the accuracy and efficiency of labeling.

[0037] The present application proposes a pixel-level convective cloud labeling system based on FY-4A: mainly composed of a software interface and a data processing module. The software interface provides a visual operation platform for the user, and the data processing module is responsible for processing and analyzing the FY-4A satellite data.

[0038] The software interface includes a main image display box and three sub-image display boxes, as shown in the accompanying Figure 1 The main image display box is used to show an example of the labeling result overlaid on the brightness temperature image, wherein the red or green area is the area actually identified as the convective cloud. The three sub-image display boxes display the radar echo image, the visible light image and the sequence brightness temperature color map from top to bottom, and the sequence brightness temperature color map is used to help the labeling personnel to more easily select the segmentation threshold. The software interface is also provided with a plurality of operation buttons, such as "select folder", "previous", "next", "switch channel", "rectangle", "polygon", "clear", "probe" and the like, to facilitate the user to perform data selection, image switching, region selection and threshold adjustment and other operations.

[0039] The data processing module is connected with the software interface through a data transmission interface, receives the operation instructions of the user on the software interface, and performs corresponding processing and analysis on the FY-4A satellite data. As shown in the accompanyingFigure 2 As shown in the figure, the data processing module mainly includes a data reading submodule, a threshold calculation submodule, and a labeling result generation submodule. The data reading submodule is responsible for reading FY-4A satellite data, the threshold calculation submodule calculates the labeling threshold of convective clouds according to the parameters set by the user and the satellite data, and the labeling result generation submodule labels the satellite data according to the calculated threshold and feeds back the labeling result to the software interface for display.

[0040] The workflow is as follows: a) Data selection and loading As shown in the figure, Figure 3 - As shown in the figure, Figure 6 As shown in the figure, the user clicks the "Select Folder" button to select the directory where the satellite data is located. Then select the date to be labeled, and then select the specific time image for labeling. The software will load the selected satellite data into the data processing module for processing.

[0041] b) Image display and operation As shown in the figure, Figure 1 As shown in the figure, the software interface displays the main image and three sub-images, the main image is an example of the labeling result overlaid on the brightness temperature image, and the sub-images are radar echo image, visible light image and sequence brightness temperature color chart. The user can switch to the images of the previous and next time through the "Previous" and "Next" buttons, and switch the main image to brightness temperature image or brightness temperature color chart through the "Switch Channel" button (shortcut key: Ctrl + s (switch)). The user can also use the "Rectangle" button to select a rectangular area, and then the three sub-images will display the corresponding area, and the threshold can be adjusted to re-segment the convective in the selected area. Click the selection box to adjust the size and position of the selection box (shortcut key: Ctrl + a (all) select the entire image).

[0042] c) Data labeling process c1. Judge the pre-segmentation result: as shown in the figure, Figure 7 As shown in the figure, first judge whether the pre-segmentation (default red part) result is convective.

[0043] c2. Delete non-convective area: as shown in the figure, Figure 8 If it is not convective, use "Rectangle" or "Polygon" to select and delete, and delete the convective that is dissipating, only keep the convective that is generating, developing and can last for a period of time.

[0044] c3. Fine-tune the threshold: as shown in the figure, Figure 9 As shown in the figure, fine-tune the threshold to separate multiple targets (multiple peaks) or remove the tail of the segmented convective and some scattered points, and only keep the main body of the convective.

[0045] In embodiments, the labeling system of the present application uses the following steps: 1. Open the labeling tool and enter the initialization page, as shown in the attached Figure 3 ; 2. Click the Select Folder button to select the data directory where the FY-4A AGRI channel data is located, and the labeling software will automatically read all the data under this folder, as shown in the attached Figure 4 ; 3. Click the Date button to select the date and time to be labeled, as shown in the attached Figure 5 ; Click the Data drop-down box to select the time to be labeled. The observation time interval of FY-4A data is 15 minutes, and a total of 96 data will be generated in a day, as shown in the attached Figure 6 ; 4. After selecting the labeling time, the software will automatically read the data to be labeled and automatically perform data preprocessing. The software will perform pre-segmentation on the satellite cloud image according to the built-in threshold table, and the pre-segmented area will be labeled as a possible convective cloud, and the pre-segmented area will be displayed as a red area in the main view, as shown in the attached Figure 7 ; 5. The labeling personnel judge whether the pre-segmented area is a convective cloud by combining visible light and brightness temperature thresholds. If it is not a convective cloud, use "rectangle" or "polygon" to select and delete it. The convective cloud that is dissipating is also deleted, and only the convective cloud that is generating, developing and still lasting for a period of time is kept, as shown in the attached Figure 8 ; Then, fine-tune the threshold to separate multiple targets (multiple peaks) or remove the tail of the segmented convective cloud and some scattered points, and only the main body of the convective cloud is kept, as shown in the attached Figure 9 ; 6. After the labeling personnel finish labeling, they can click the "Next" button to automatically read the next time's data to be labeled and display it in the labeling interface.

[0046] The judgment method of convective cloud labeling is as follows: (1) The most intuitive manifestation of convective cloud is that it will bring a certain degree of precipitation, so the radar echo intensity of the corresponding area is an effective means to test the convective cloud. Therefore, the present application selects the area of 16°48'N-57°4'N and 94°79'E-161°34'E in the full disc area. The area within this range has national radar stitching data, so it is used as a labeling verification means.

[0047] (2) The manifestation of convective cloud on the brightness temperature channel image is dense and white, and the manifestation on the visible light channel is more intuitive. Convective cloud and cirrus water vapor have obvious differences on the visible light channel. Convective cloud has poor light transmission and obvious bubble shape, while cirrus and water vapor clouds look more loose and smooth, with poor texture.

[0048] (3) The brightness temperature of convective clouds shows great differences in different regions. In the southern region, the brightness temperature threshold of convective clouds is lower, generally between 215K and 225K, but in the northern region, it is generally between 220K and 235K, and even up to about 240K. In terms of radar echo intensity, the radar echo intensity of convective clouds is basically greater than 25dbz.

[0049] (4) There are obvious seasonal differences in convective clouds. In January, February, March, October, November and December, convective clouds are basically distributed in the region south of the Qinling and Huaihe rivers, and the cloud clusters in the northern region are basically high-level cirrus clouds. In the remaining months, convective clouds will be distributed in both northern and southern regions, mainly concentrated in the northwest, north China and northeast regions, and large-scale complex cloud systems are prone to occur in the northeast and north China regions, and the distribution in the southern region is more uniform.

[0050] (5) When annotating convective clouds, try to distinguish different convective cloud clusters with different brightness temperature peaks from cirrus and water vapor, so as to have obvious boundaries. The distinction mainly depends on the adjustment of the brightness temperature threshold, and the temperature of convective clouds is lower than that of cirrus or water vapor, and the temperature of different convective clouds also has different differences according to the region. The distinction logic is shown in the accompanying Figure 10

[0051] (6) Convective clouds have a time sequence in the development process, and the continuity of the same convective cloud cluster should be ensured as much as possible during data annotation. When annotating, the threshold selection of the same convective cloud target between adjacent frames should not have great differences, and the change range should be maintained between 2K and 5K, and the specific value is determined according to the development or dissipation trend of convective clouds. The final annotation result is shown in the accompanying Figure 11

[0052] An electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0053] A computer readable storage medium for storing computer instructions, the computer instructions are executed by a processor to implement the steps of the above method.

[0054] ​​The memory in the embodiments of the application can be a volatile memory or a nonvolatile memory, or can include both volatile and nonvolatile memory. Where the nonvolatile memory is a read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), electrically EPROM (EEPROM), or flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example, and not limitation, many forms of RAM are available, for example, static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM). Note that the memory described herein is intended to include, among others, these and any other memory suitable for storing the data adaptively described herein.

[0055] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable apparatus. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transferred from one website, computer, server or data center to another website, computer, server or data center through wired, such as coaxial cable, optical fiber, digital subscriber line (DSL) or wireless, such as infrared, wireless, microwave, etc. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be magnetic media, such as floppy disks, hard disks, magnetic tapes, optical media, such as digital video discs (DVD), or semiconductor media, such as solid state discs (SSD), etc.

[0056] In the implementation process, each step of the above method can be completed by integrated logic circuit of hardware in the processor or instruction in the form of software. The steps of the method disclosed in the embodiments of the present application can be directly embodied as hardware processor execution or executed by combination of hardware and software modules in the processor. The software module can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, register or other mature storage medium in the art. The storage medium is located in the memory, and the processor reads the information in the memory and combines the hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0057] It should be noted that the processor in the embodiments of the present application can be an integrated circuit chip with signal processing capability. In the implementation process, each step of the method embodiments can be completed by integrated logic circuits or instructions in the form of software in the processor. The processor can be a general processor, a digital signal processor DSP, an application specific integrated circuit ASIC, a field programmable gate array FPGA or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The disclosed methods, steps and logic block diagrams in the embodiments of the present application can be implemented or executed. The general processor can be a microprocessor or the processor can be any conventional processor. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as hardware code processor execution completion, or executed by hardware and software module combination in the code processor. The software module can be located in the random access memory, the flash memory, the read only memory, the programmable read only memory or the electrically erasable programmable memory, the register or other mature storage medium in the art. The storage medium is located in the memory, and the processor reads the information in the memory, and combines the hardware to complete the steps of the above method.

[0058] The pixel-level convection cloud annotation system based on FY-4A proposed in the present application is described in detail, the principle and implementation of the present application are described, the above embodiment is only used to help understand the method of the present application and its core idea; at the same time, for the general technical personnel in the art, according to the idea of the present application, the specific implementation and application range will be changed, according to the above, the content of the specification should not be understood as the limitation of the present application.

Claims

1. A pixel-level convection cloud annotation system based on FY-4A, characterized in that: the system comprises a software interface and a data processing module; the software interface provides a visual operation platform for users, including a main image display box and three sub-image display boxes, the main image display box is used to show an example of annotation results overlaid on a brightness temperature image, and the three sub-image display boxes sequentially display a radar echo image, a visible light image and a sequence brightness temperature color map from top to bottom; the data processing module is connected with the software interface through a data transmission interface, and processes and analyzes FY-4A satellite data, including a data reading submodule, a threshold calculation submodule and an annotation result generation submodule; the data reading submodule is used to read FY-4A satellite data; the threshold calculation submodule calculates the annotation threshold of convection cloud according to the parameters set by the user and the satellite data; the annotation result generation submodule annotates the satellite data according to the calculated threshold, and feeds back the annotation results to the software interface for display. 2.The system according to claim 1, characterized in that: the sequence brightness temperature color map is used to assist the annotation personnel in selecting a segmentation threshold. 3.The system according to claim 2, characterized in that: the software interface further comprises an operation button group, specifically: a data selection button, an image switching button, a region selection button and a threshold adjustment button; used for selecting a folder, the previous image, the next image, switching channels, a rectangle, a polygon, clearing and a probe operation. 4.The system according to claim 3, characterized in that: the data annotation process of the data processing module is as follows: pre-segmentation: generating a red pre-annotation region according to the built-in threshold table when loading data; deleting non-convection regions: the user selects non-convection clouds or dissipating state convection clouds in the pre-segmentation through the region selection button, and deletes them; fine-tuning the threshold: adjusting the threshold to separate multi-peak targets or remove scattered points, and retaining the main body of the convection cloud. 5.The system according to claim 4, characterized in that: in the data annotation process, the threshold selection variation of the same convection cloud target between adjacent frames is maintained between 2K-5K. 6.The system according to claim 5, characterized in that: the convection cloud annotation threshold in the data annotation is divided by region: southern region: 215K-225K; northern region: 220K-240K. 7.The system according to claim 6, characterized in that: in the data annotation, different convection cloud groups and cirrus water vapor are distinguished by adjusting the brightness temperature threshold, and the temperature of the convection cloud is lower than that of the cirrus or water vapor cloud group; the temperature of different convection clouds also has different differences according to the region. 8.An annotation method for the pixel-level convection cloud annotation system based on FY-4A according to any one of claims 1 to 7, characterized in that: the method specifically comprises the following steps: step 1: loading FY-4A AGRI data through the software interface, automatically performing pre-segmentation and generating an initial annotation region in the main image display box; Step 2: Check the initial labeling area based on the radar echo image, visible light image and sequence brightness temperature color map of the sub-image display frame, and delete non-convective clouds and dissipating convective clouds; Step 3: Fine-tune the brightness temperature threshold to separate multi-peak targets and remove scattered labeling points; Step 4: Generate pixel-level convective cloud labeling results. 9.An electronic device comprising a memory and a processor, the memory storing a computer program, wherein, The processor executes the computer program to realize the steps of the method in claim 8.

10. A computer readable storage medium for storing computer instructions, characterized in that, The computer instructions are executed by the processor to realize the steps of the method in claim 8.

Images