Search

CN-121811277-B - Wildfire smoke group and boundary vector automatic identification method, system, storage medium and electronic equipment thereof

CN121811277BCN 121811277 BCN121811277 BCN 121811277BCN-121811277-B

Abstract

The invention discloses a wildfire smoke group and a boundary vector automatic identification method, a system, a storage medium and electronic equipment thereof, belonging to the technical field of satellite remote sensing and disaster monitoring. The method comprises the steps of obtaining satellite active fire points, fire events and UVAI data, positioning a fire source area based on the geographical position of the fire event and marking initial smoke plume pixels, identifying a smoke group connected domain with continuous space through an iterative area growing algorithm, automatically generating an accurate closed boundary vector by adopting a matched geometric algorithm according to the specific distribution form of the connected domain pixels in a satellite pixel matrix, and finally screening independent smoke groups with isolated space by utilizing multiple criteria. The invention avoids the defect that the traditional image method is easy to be interfered by environment in physical principle, realizes the full-automatic treatment without training data, can output vector boundaries which can be directly used for diffusion and emission evaluation, and remarkably improves the accuracy, the degree of automation and the application value of monitoring the wildfire smoke group.

Inventors

  • LIU NAIAN
  • FU YUYUN
  • XIE XIAODONG

Assignees

  • 中国科学技术大学

Dates

Publication Date
20260508
Application Date
20260311

Claims (10)

  1. 1. The automatic identification method for the wildfire smoke group and the boundary vector thereof is characterized by comprising the following steps: S1, acquiring satellite active fire point data, fire event information and track-by-track absorption aerosol index UVAI data of a target area and a target period; s2, aiming at each wild fire event, calculating the central longitude and latitude of the fire passing area according to the fire event information, and defining a fire source area according to the central longitude and latitude, traversing the track UVAI data from the beginning of a fire to the period of extinction, and marking UVAI pixels in the fire source area as smoke plume pixels on the track which meets the requirement of the track passing on the day that the satellite is detected to be active in the fire event combustion boundary; S3, aiming at each orbit, using the smoke plume pixels as starting pixels, searching adjacent pixels according to the adjacent relation of row numbers +/-1 and column numbers +/-1 based on row numbers and column numbers of the pixels in a satellite orbit pixel matrix, and iteratively combining the adjacent pixels meeting a threshold condition into the smoke plume pixels according to a preset UVAI minimum threshold until no adjacent pixels meeting the threshold condition are newly added, so as to obtain a wildfire event smoke mass connected domain; S4, identifying the boundary of the smoke group connected domain and forming a corresponding closed boundary vector of the smoke group connected domain by track according to the spatial distribution form of UVAI pixels in the smoke group connected domain in the track pixel matrix for each track; S5, based on the track-by-track smoke group connected domain and the closed boundary vector thereof, setting discrimination conditions to screen the independent smoke groups of the wild fires which are distributed in a space isolated mode.
  2. 2. The method for automatically identifying a wildfire smoke group and a boundary vector thereof according to claim 1, wherein in step S1, the satellite active fire point data at least comprises longitude and latitude and detection time information of the active fire point; The fire event information at least comprises the longitude and latitude of a combustion boundary and the fire starting time of each wildfire event fire passing area; The UVAI data at least comprises a pixel UVAI value, pixel center longitudes and latitudes and longitudes of four vertexes of a rectangular area covered by the pixel.
  3. 3. The method for automatically identifying the cigarette group and the boundary vector thereof according to claim 1, wherein in the step S2, the longitude and latitude of the center of the over-fire area are obtained by averaging the longitude and latitude of all boundary points of the over-fire area; the range of the fire source area is determined by UVAI pixels of the fire source area, and the number is 1-10.
  4. 4. The method for automatically recognizing a wildfire smoke and a boundary vector thereof according to claim 1, wherein in step S2, for given track-by-track UVAI data, it is determined whether a satellite active fire point is detected within the fire event combustion boundary on the day of the passing of the track; and marking the picture element of the fire source area UVAI as a smoke plume picture element for the orbit containing the satellite active fire points.
  5. 5. The method of automatic identification of a group of wildfires and their boundary vectors according to claim 1, wherein the iterative merging in step S3 comprises: and repeatedly executing the cycle of identifying the adjacent pixels of the marked smoke plume pixels and carrying out the minimum threshold comparison between UVAI and the adjacent pixels, marking the adjacent pixels meeting the threshold condition as the smoke plume pixels and merging until the adjacent pixels cannot meet the UVAI minimum threshold condition.
  6. 6. The method for automatically identifying the smoke mass and the boundary vector thereof according to claim 1, wherein in the step S4, when the smoke mass connected domain only comprises 1 UVAI pixels, the longitudes and latitudes of four vertexes of the pixels are extracted and sequentially stored according to the original sequence of the four vertexes to form the boundary vector of the smoke mass connected domain; When the smoke group communicating domain contains no less than 2 UVAI pixels, determining boundary points of the smoke group communicating domain based on the longitude and latitude of the vertexes of the rectangular area covered by each UVAI pixel of the smoke group communicating domain, and sequentially storing the boundary points according to a preset sequence to form a closed boundary vector of the smoke group communicating domain.
  7. 7. The automatic identification method for a wildfire smoke group and boundary vectors thereof according to claim 1, wherein in step S5, the discrimination conditions at least include: (a) Judging whether the smoke group communicating domain is positioned at the orbit edge of the satellite UVAI according to the maximum and minimum row numbers and the column numbers of the smoke group communicating domain pixels, and screening smoke groups at the non-orbit edge; (b) Based on the maximum value of the smoke group connected domain pixels UVAI, the average value of the smoke group connected domain pixels UVAI, the average value of the background UVAI and the satellite active fire number NOF bg within 20km of the boundary extension of the smoke group during fire, comparing the NOF bg with a corresponding threshold THR_max smoke 、THR_mean smoke 、THR_mean bg 、THR_NOF bg to screen the wild fire smoke groups in isolated space distribution; (c) Setting a region range by taking the longitude and latitude of the center of the fire passing region in the step S2 as the center, and screening a smoke mass connected region smaller than the region range.
  8. 8. The automatic identification system for the wildfire smoke mass and the boundary vector thereof for realizing the method of any one of claims 1 to 7 is characterized by comprising a data acquisition unit, a central longitude and latitude calculation and fire source region demarcation unit, a smoke plume pixel marking unit, a connected domain identification unit, a boundary vector generation unit and an independent smoke mass screening unit; The data acquisition unit is respectively connected with the central longitude and latitude calculation and fire source region demarcation unit and the smoke plume pixel marking unit and is used for outputting fire event information, satellite active fire point data and track-by-track UVAI data; The center longitude and latitude calculation and fire source region demarcating unit is connected with the smoke plume pixel marking unit and is used for outputting the center longitude and latitude of the fire passing region and the fire source region; the smoke plume pixel marking unit, the communicating region identifying unit and the boundary vector generating unit are sequentially connected and are used for sequentially outputting smoke plume pixels, smoke mass communicating regions and closed boundary vectors of the smoke mass communicating regions track by track; the independent smoke group screening unit is connected with the central longitude and latitude calculation and fire source region demarcation unit, the connected region identification unit and the boundary vector generation unit and is used for screening the independent smoke groups of the wild fires with isolated space distribution based on rail edge discrimination, UVAI statistics, background statistics, threshold discrimination of the number of active fire points and region range discrimination.
  9. 9. A computer readable storage medium having stored thereon a computer program which, when executed by a processor, causes the processor to perform the method of any of claims 1 to 7.
  10. 10. An electronic device comprising a processor and a memory, the memory having stored therein a computer program which, when executed by the processor, causes the processor to perform the method of any of claims 1 to 7.

Description

Wildfire smoke group and boundary vector automatic identification method, system, storage medium and electronic equipment thereof Technical Field The invention relates to the technical field of satellite remote sensing and disaster monitoring, in particular to a method, a system, a storage medium and electronic equipment for realizing automatic identification of independent smoke groups of wild fires and accurate boundary vector extraction by utilizing satellite ultraviolet band absorption aerosol indexes (UVAI) and combining active fire points and fire event information. Background Wildfires act as an important natural source of emissions of aerosols, greenhouse gases and pollutants in the earth's atmosphere, and the diffusion and transport of the resulting smoke mass poses a serious threat to air quality and public health in downwind areas. Therefore, the accurate identification, space morphology definition and dynamic tracking of the independent smoke mass of the wild fire are important preconditions for pollution early warning, exposure risk assessment and emission quantification. However, the current mainstream wildfire smoke recognition technology has a plurality of key bottlenecks, which restrict the application of the wildfire smoke recognition technology in actual business monitoring and fine scientific research. The prior art relies on satellites or video images based on the visible light band and uses traditional image processing models (such as RGB color models) or artificial intelligence methods for smoke segmentation. The method is firstly obviously restricted by environmental conditions, and the identification performance is easily influenced by illumination intensity, shooting angle, cloud cover and background interference, so that the reliability is reduced under complex meteorological conditions. More fundamentally, it is difficult for visible light characteristics to effectively distinguish smoke with similar spectral and textural characteristics from thin clouds, dust or city haze, thereby producing a high false positive rate. Secondly, although the method based on the artificial intelligent model such as deep learning is excellent in some scenes, the method relies on a large amount of high-quality marked smoke image data for training, and the method for acquiring large-range and diversified wild fire smoke group marking data is high in cost and huge in challenge, and meanwhile, a large amount of computing resources are generally consumed in the model training and deducing process. Finally, and most critical, existing approaches generally focus on pixel-level identification of "smoke regions," and have difficulty effectively correlating and stripping out complete individual smoke mass individuals from a particular fire event in time and space. The smoke clusters generated by different fire sources or the smoke plumes of the same fire source in different diffusion stages cannot be automatically distinguished, and the capability of automatically extracting the accurate two-dimensional space boundary of the smoke clusters is further lacking, so that quantitative analysis on the diffusion path, the morphological evolution and the corresponding emission amount of the single smoke clusters is difficult to support. Therefore, a solution that is more reliable in principle, has high automation degree, and can output the accurate vector space shape of the independent smoke mass is needed in the industry, so as to break through the limitation of the current technology. Disclosure of Invention The invention aims to solve the technical problems that the wild fire smoke group identification method in the prior art is weak in anti-interference capability, depends on training data, and cannot automatically output an accurate vector boundary of an independent smoke group associated with a specific fire disaster event. In order to solve the above problems, in one aspect, the present invention provides a method for automatically identifying a wildfire smoke group and a boundary vector thereof, including: S1, acquiring satellite active fire point data, fire event information and track-by-track absorption aerosol index UVAI data of a target area and a target period; s2, aiming at each wild fire event, calculating the central longitude and latitude of the fire passing area according to the fire event information, and defining a fire source area according to the central longitude and latitude, traversing the track UVAI data from the beginning of a fire to the period of extinction, and marking UVAI pixels in the fire source area as smoke plume pixels on the track which meets the requirement of the track passing on the day that the satellite is detected to be active in the fire event combustion boundary; S3, aiming at each orbit, using the smoke plume pixels as starting pixels, searching adjacent pixels according to the adjacent relation of row numbers +/-1 and column numbers +/-1 based on row numbers and column