CN-116189007-B - Fire point determining method, fire point determining device, electronic equipment and computer readable storage medium

CN116189007BCN 116189007 BCN116189007 BCN 116189007BCN-116189007-B

Abstract

The invention discloses a fire point determining method, a fire point determining device, electronic equipment and a computer readable storage medium. The method comprises the steps of obtaining a target image of a target place and a historical fire point multi-dimensional probability image corresponding to the target place, obtaining an initial fire point probability image according to the target image, wherein the initial fire point probability image is an image which displays initial probability of occurrence of a fire point at a position corresponding to a pixel point in the target image, and obtaining the target fire point probability image according to the target image, the historical fire point multi-dimensional probability image and the initial fire point probability image, wherein the target fire point probability image is an image which displays target probability of occurrence of the fire point at the position corresponding to the pixel point in the target image. The invention solves the technical problem that the determined fire probability is inaccurate when the fire probability of a certain position is determined in the related technology.

Inventors

ZHANG RUIZHE
Tian Chenyu
ZHOU KAI
WANG YANI
YE KUAN
LI HONGDA
CAI YINGMIAO
LI CHUNSHENG
WU LEI
YIN ZHIPING

Assignees

国网北京市电力公司
国家电网有限公司
北京深蓝空间遥感技术有限公司

Dates

Publication Date: 20260508
Application Date: 20230228

Claims (9)

1. A fire point determination method, comprising: acquiring a target image of a target place and a historical fire point multi-dimensional probability image corresponding to the target place, wherein the historical fire point multi-dimensional probability image is an image which displays the historical probability that the fire points appear in a plurality of dimensions at the positions corresponding to pixel points in the image, the historical fire point multi-dimensional probability image is obtained according to the historical data of the target place, and the historical data comprises the historical fire point positions, the occurrence times and the occurrence seasons; Obtaining an initial fire point probability image according to the target image, wherein the initial fire point probability image is an image which displays initial probability of occurrence of fire points at positions corresponding to pixel points in the target image; obtaining a target fire point probability image according to the target image, the historical fire point multi-dimensional probability image and the initial fire point probability image, wherein the target fire point probability image is an image which displays the target probability of the fire point at the position corresponding to the pixel point in the target image; The method comprises two-dimension and normalization of data corresponding to the target image, and calculation of autocorrelation matrix by autocorrelation function Selecting a pixel point with the brightness value of the first 1% in the target image as an initial fire point to obtain an initial fire point data set, determining a clustering center position of the initial fire point data set through k-means clustering, and using a numerical value corresponding to the clustering center position As a reference fire point, according to linear filter parameters And carrying out linear weighted summation on each pixel point in the target image to obtain the probability of fire points at the position corresponding to each pixel point in the initial fire point probability image, wherein the linear filter parameter is calculated according to the following formula: 。
2. The method of claim 1, wherein the deriving a target fire probability image from the target image, the historical fire multi-dimensional probability image, and the initial fire probability image comprises: The target image, the historical fire point multi-dimensional probability image and the initial fire point probability image are input to a target fire point detection model to obtain the target fire point probability image, wherein the target fire point detection model is obtained by training the initial fire point detection model according to sample data, the sample data comprise a positive sample image, the historical positive sample fire point multi-dimensional probability image, the initial positive sample fire point probability image, the target positive sample fire point probability image and a negative sample image, the historical negative sample fire point multi-dimensional probability image, the initial negative sample fire point probability image and the target negative sample fire point probability image, the target positive sample fire point probability image is an image, the probability of occurrence of a fire point at a position corresponding to a pixel point in the image is larger than a first threshold, the target negative sample fire point probability image is an image, the probability of occurrence of a fire point at a position corresponding to the pixel point in the image is smaller than a second threshold, and the first threshold is larger than the second threshold.
3. The method of claim 2, wherein the inputting the target image, the historical fire multi-dimensional probability image, and the initial fire probability image to a target fire detection model, prior to obtaining the target fire probability image, further comprises: acquiring an initial fire detection model, and constructing a loss function for model training, wherein the loss function comprises a cross entropy loss function; and training the initial fire point detection model by adopting the sample data based on the loss function to obtain the target fire point detection model.
4. The method of claim 1, wherein the deriving a target fire probability image from the target image, the historical fire multi-dimensional probability image, and the initial fire probability image comprises: Determining a probability clustering result of fire points at positions corresponding to pixel points in the target image according to the historical fire point multi-dimensional probability image and the initial fire point probability image; and obtaining the target fire point probability image according to a probability clustering result of the fire points at the positions corresponding to the pixel points in the target image.
5. The method of claim 1, wherein prior to the acquiring the target image of the target site, further comprising: Acquiring an initial image of the target site; Registering the initial image according to preset parameter information to obtain the target image, wherein the preset parameter information comprises preset resolution and spatial corresponding position information of each vertex of the image.
6. The method of claim 1, wherein obtaining a historical fire multi-dimensional probability image corresponding to the target location comprises: acquiring historical fire point data of the target site; And determining the multi-dimensional probability image of the historical fire point corresponding to the target place according to the historical fire point data, wherein the number of the multi-dimensional dimensions is determined according to the type number of the historical fire point data.
7. A fire point determining apparatus, characterized by comprising: The system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring a target image of a target place and a historical fire point multi-dimensional probability image corresponding to the target place, the historical fire point multi-dimensional probability image is an image which displays the historical probability that fire points appear on a plurality of dimensions at positions corresponding to pixel points in the image, the historical fire point multi-dimensional probability image is obtained according to historical data of the target place, and the historical data comprises historical fire point positions, occurrence times and occurrence seasons; the second acquisition module is used for obtaining an initial fire point probability image according to the target image, wherein the initial fire point probability image is an image which displays initial probability of occurrence of fire points at positions corresponding to pixel points in the target image; the third acquisition module is used for acquiring a target fire point probability image according to the target image, the historical fire point multi-dimensional probability image and the initial fire point probability image, wherein the target fire point probability image is an image which displays the target probability of the fire point at the position corresponding to the pixel point in the target image; The method comprises two-dimension and normalization of data corresponding to the target image, and calculation of autocorrelation matrix by autocorrelation function Selecting a pixel point with the brightness value of the first 1% in the target image as an initial fire point to obtain an initial fire point data set, determining a clustering center position of the initial fire point data set through k-means clustering, and using a numerical value corresponding to the clustering center position As a reference fire point, according to linear filter parameters And carrying out linear weighted summation on each pixel point in the target image to obtain the probability of fire points at the position corresponding to each pixel point in the initial fire point probability image, wherein the linear filter parameter is calculated according to the following formula: 。
8. An electronic device, comprising: A processor; a memory for storing the processor-executable instructions; wherein the processor is configured to execute the instructions to implement the fire determination method of any one of claims 1 to 6.
9. A computer readable storage medium, characterized in that instructions in the computer readable storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the fire determination method according to any one of claims 1 to 6.

Description

Fire point determining method, fire point determining device, electronic equipment and computer readable storage medium Technical Field The present invention relates to the field of detection, and in particular, to a fire point determining method, a fire point determining device, an electronic device, and a computer readable storage medium. Background The fire point detection is a technology for observing a large-range area by utilizing a remote sensing technology and determining the ignition position by utilizing a data intelligent analysis technology, and has important significance for power grid safety, forest protection and the like. However, in the related art, when determining the fire probability of a certain position, there is a problem that the determined fire probability is inaccurate. In view of the above problems, no effective solution has been proposed at present. Disclosure of Invention The embodiment of the invention provides a fire point determining method, a fire point determining device, electronic equipment and a computer readable storage medium, which are used for at least solving the technical problem that the determined fire point probability is inaccurate when the fire point probability of a certain position is determined in the related technology. According to one aspect of the embodiment of the invention, a fire point determining method is provided, which comprises the steps of obtaining a target image of a target place and a historical fire point multi-dimensional probability image corresponding to the target place, wherein the historical fire point multi-dimensional probability image is an image which displays historical probabilities that fire points appear in multiple dimensions at corresponding positions of pixels in the image, obtaining an initial fire point probability image according to the target image, wherein the initial fire point probability image is an image which displays initial probabilities that fire points appear at corresponding positions of pixels in the target image, and obtaining a target fire point probability image according to the target image, the historical fire point multi-dimensional probability image and the initial fire point probability image, wherein the target fire point probability image is an image which displays target probabilities that fire points appear at corresponding positions of pixels in the target image. Optionally, the obtaining a target fire probability image according to the target image, the historical fire multi-dimensional probability image and the initial fire probability image comprises inputting the target image, the historical fire multi-dimensional probability image and the initial fire probability image into a target fire detection model to obtain the target fire probability image, wherein the target fire detection model is obtained by training an initial fire detection model according to sample data, the sample data comprises a positive sample image, a historical positive sample fire multi-dimensional probability image, an initial positive sample fire probability image, a target positive sample fire probability image and a negative sample image, the historical negative sample fire multi-dimensional probability image, the initial negative sample fire probability image and the target negative sample fire probability image, the target positive sample fire probability image is an image in which the probability of occurrence of a fire at a pixel point corresponding position in the image is larger than a first threshold, and the target negative sample fire probability image is an image in which the probability of occurrence of a fire at a pixel point corresponding position in the image is smaller than a second threshold, and the probability of occurrence of the fire at the pixel point corresponding position in the image is larger than the second threshold. Optionally, the method comprises the steps of inputting the target image, the historical fire point multi-dimensional probability image and the initial fire point probability image to a target fire point detection model, acquiring the initial fire point detection model and constructing a loss function for model training before obtaining the target fire point probability image, wherein the loss function comprises a cross entropy loss function, training the initial fire point detection model by adopting the sample based on the loss function, and obtaining the target fire point detection model. Optionally, the obtaining a target fire point probability image according to the target image, the historical fire point multi-dimensional probability image and the initial fire point probability image comprises determining a probability clustering result of fire points at positions corresponding to pixels in the target image according to the historical fire point multi-dimensional probability image and the initial fire point probability image, and obtaining the target fire point probabilit