Forest fire monitoring method and system based on remote sensing data, and medium

Abstract

The application discloses a forest fire monitoring method and system based on remote sensing data and a medium; an optimal fire point sensitive band group is acquired based on historical fire data, a brightness temperature feature representation is determined under the optimal fire point sensitive band group, a fire point monitoring index function about a low-light band brightness and a vegetation index is constructed, and forest fire point monitoring is performed after the fire point monitoring index function is corrected by using the brightness temperature feature representation; the application improves the fire point identification method on the basis of the existing forest fire monitoring mechanism, constructs the fire point monitoring index function in combination with historical fire data and remote sensing data, considers the low-light band brightness and the vegetation index in the fire point monitoring index function to distinguish forest fires from urban combustion heat sources, and corrects the fire point monitoring index function based on the brightness temperature feature representation to distinguish fire edge positions from urban combustion heat sources, so that the identification of forest fires has sensitivity, stability and accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fire monitoring, in particular to a forest fire monitoring method and system based on remote sensing data and a medium. BACKGROUND

[0002] Forest resources are scarce, and once a forest fire occurs, large areas of forest will be reduced to ashes, causing serious losses. At the same time, forest land is prone to soil erosion, prone to water and flood disasters, and even affects the ecological environment and sustainable development of the entire region.

[0003] At present, the most effective means is to use remote sensing technology to monitor and analyze the resources in the forest area in a wide range, a wide field of view, and 24 hours a day, and to automatically alarm when a fire or smoke is found in the forest area, and to locate the position of the fire point. Once a fire occurs, an advanced and reasonable monitoring mechanism can automatically alarm and timely notify relevant personnel.

[0004] However, the existing forest fire monitoring mechanism still has many shortcomings: on the remote sensing image, it is difficult to simply rely on radiation brightness or temperature to fully separate forest fires from urban combustion heat sources, resulting in low forest fire monitoring accuracy and false alarms. SUMMARY

[0005] The technical problem to be solved by the present application is that the existing forest fire monitoring mechanism on the remote sensing image is difficult to simply rely on radiation brightness or temperature to fully separate forest fires from urban combustion heat sources, which results in low forest fire monitoring accuracy and false alarms. The present application aims to provide a forest fire monitoring method, system and medium based on remote sensing data, which combines historical fire data and remote sensing data to construct a fire point monitoring index function. On the one hand, the fire point monitoring index function considers the micro-light band brightness and vegetation index to distinguish forest fires from urban combustion heat sources. On the other hand, the fire point monitoring index function is corrected based on the brightness temperature feature representation to distinguish fire edges from urban combustion heat sources, achieving sensitive, stable and accurate identification of forest fires.

[0006] The present application is achieved by the following technical solutions:

[0007] The present application provides a forest fire monitoring method based on remote sensing data, comprising:

[0008] S1, obtaining historical fire data of a forest area to be monitored, the historical fire data comprising fire point data and remote sensing data;

[0009] S2, constructing an optimal fire point sensitive band group based on the historical fire data: combining bands based on band characteristics and inter-band correlation coefficients, and taking the combination with the maximum OIF index as the optimal fire point sensitive band group;

[0010] S3, determining a brightness temperature characteristic representation under the optimal fire point sensitive wave band group;

[0011] S4, constructing a fire point monitoring index function about the micro-light wave band brightness and the vegetation index, and correcting the fire point monitoring index function based on the brightness temperature characteristic representation;

[0012] S5, inputting real-time remote sensing data of a forest area to be monitored into the corrected fire point monitoring index function to monitor the forest fire point.

[0013] The working principle of the scheme is: the existing forest fire monitoring mechanism is difficult to fully separate the forest fire from the urban combustion heat source on the remote sensing image by simply relying on the radiation brightness or temperature, fully separates the forest fire from the urban combustion heat source, and leads to low forest fire monitoring precision and false alarms; the present application aims to provide a forest fire monitoring method based on remote sensing data, improve the fire point recognition method on the existing forest fire monitoring mechanism, and construct a fire point monitoring index function combining historical fire data and remote sensing data; on the one hand, the fire point monitoring index function considers the micro-light wave band brightness and the vegetation index to distinguish the forest fire from the urban combustion heat source; on the other hand, the fire point monitoring index function is corrected based on the brightness temperature characteristic representation to distinguish the fire edge position from the urban combustion heat source, so as to realize the recognition of the forest fire with sensitivity, stability and accuracy.

[0014] In the remote sensing data, the forest fire is manifested as an increase in the micro-light wave band brightness and the ground object temperature, but the forest fire and the urban heat source such as industrial combustion and urban heat island are confused in the radiation brightness and the ground object temperature, and it is difficult to fully separate the forest fire from other heat sources by simply relying on the micro-light wave band brightness or the ground object temperature; therefore, the present application constructs a fire point monitoring index function about the micro-light wave band brightness and the vegetation index, increases the difference between the forest heat source area and the urban heat source area through the vegetation index; however, the radiation brightness and the ground object temperature of the micro-light wave band are reduced at the fire point edge, and the edge suspected fire pixels and the urban center pixels are also prone to confusion, so the present application corrects the fire point monitoring index function based on the brightness temperature characteristic representation to increase the recognition degree of the edge suspected fire position and the urban center position.

[0015] Further optimization scheme is that step S2 includes the following substeps:

[0016] S21, acquiring fire point data, and extracting non-fire point data corresponding to the fire point data in space from the remote sensing data;

[0017] S22, calculating characteristic values of the fire point data and the non-fire point data in each wave band, the characteristic values including maximum value, minimum value, average value and standard deviation; finding a wave band with the largest difference between the characteristic values of the fire point data and the non-fire point data as a basic wave band;

[0018] S23, calculate the correlation coefficient between each band and the basic band, and screen out bands with correlation coefficients less than the correlation threshold Q. A The bands are selected as the first pre-selected bands. The correlation coefficients between each first-selected band are calculated, and bands with correlation coefficients less than the correlation threshold Q are selected. A The bands are selected as the second pre-selected bands; the correlation coefficient is an important basis for measuring the linear correlation between two variables; the larger the correlation coefficient, the stronger the correlation between the variables, and the more likely the information contained is to be repeated. In order to reduce data redundancy, bands with small correlation coefficients are selected first when selecting band groups.

[0019] S24, combine the basic band with the second pre-selected band: combine the basic band and the two second pre-selected bands into a band group;

[0020] S25, calculate the OIF index of each band group, and select the band group with the largest OIF index as the optimal fire-sensitive band group.

[0021] A further optimization scheme is that the correlation coefficient is calculated according to the following formula.

[0022] Q=(δ ij ) 2 / (δ ii +δ jj )

[0023] Among them, (δ ij ) 2 Let δ be the covariance between band i and band j. ii Let δ be the standard deviation of band i. jj Standard deviation of band j.

[0024] A further optimized solution is that the method for obtaining the basic band includes:

[0025] Calculate the differences between fire point data and non-fire point data: the difference between the maximum value, the difference between the minimum value, the difference between the mean value and the difference between the standard deviation;

[0026] Calculate the arithmetic mean of the differences between the maximum values, minimum values, average values, and standard deviations, and select the band with the largest arithmetic mean as the basic band.

[0027] A further optimized scheme is that the brightness temperature feature representation includes the brightness temperature representation of the current position and the brightness temperature difference representation between the current position and the background pixel;

[0028] The brightness temperature at the current location is expressed as the brightness temperature T at the current location in the fundamental band. e ;

[0029] The brightness temperature difference between the current position and the background pixel is expressed as:

[0030]

[0031] wherein, T1 is the brightness temperature of the current position in the first wave band, T2 is the brightness temperature of the current position in the second wave band, T e0 is the average value of the basic wave band background pixel temperature, the first wave band and the second wave band belong to the optimal fire point sensitive wave band group.

[0032] A further optimization scheme is that the fire point monitoring index function is:

[0033]

[0034] wherein k is the fire point monitoring index, NTL G is the normalized faint light wave band brightness, the normalization linear function normalization processing makes the value range between 0 and 1; NDVI p is the vegetation index before p time, and a is the fire point monitoring coefficient.

[0035] Considering the relationship between vegetation and human activities, the urban center usually has less vegetation coverage with the increase of building density, and the vegetation coverage increases with the decrease of building density in the suburbs away from the city, so that the high faint light wave band brightness feature is smaller in the city center and larger in the fire center, enhancing the spectral difference between forest fire and urban heat source.

[0036] A further optimization scheme is that the correction method of the fire point monitoring index function includes:

[0037] The fire point monitoring index considering the brightness temperature feature is:

[0038]

[0039] wherein T eG is the normalized current position brightness temperature representation, and ΔT is the brightness temperature difference representation of the current position and the background pixel; the normalization linear function normalization processing makes the value range between 0 and 1.

[0040] When the fire point appears, the ground object temperature will abnormally increase (represented as brightness temperature increase in remote sensing data), and the vegetation index decreases, resulting in the deviation of the brightness temperature and vegetation index ratio from the natural change range. Therefore, based on the principle that the latent heat transfer decreases with the increase of vegetation density, the fire point monitoring index is represented in the form of faint light wave band brightness and brightness temperature and vegetation index ratio.

[0041] The brightness and brightness temperature of the micro-light waveband of the fire point edge are reduced, and the edge suspected fire point pixel and the city center pixel are prone to confusion. The present scheme corrects the fire point monitoring index function by using the brightness temperature difference between the fire and the background pixel, and further improves the forest fire identification capability.

[0042] Further optimization scheme is that S5 comprises the following sub-steps:

[0043] S51, divide the map of the forest area to be monitored into N*N grids, and in each grid, select the positions of the four corners and the center of the grid as monitoring points;

[0044] S52, extract the determination data of each monitoring point from the real-time remote sensing data of the forest area to be monitored: the brightness NTL of the micro-light waveband, the vegetation index NDVI before the p time, and the brightness temperature characteristic representation of the monitoring point under the optimal fire point sensitive waveband group; p

[0045] S52, input the determination data of each monitoring point into the fire point monitoring model to calculate the fire point monitoring index k;

[0046] S53, judge whether the fire point monitoring index k exceeds the fire point threshold K, if yes, determine that the current monitoring point is a fire point; otherwise, determine that the current monitoring point is a safe point;

[0047] S54, display each fire point on the grid in different forms.

[0048] Indicated by different colors or connected together to form an equal index line. It is convenient to find the center position of the fire point in time.

[0049] The present scheme also provides a forest fire monitoring system based on remote sensing data, which is used to realize the above-mentioned forest fire monitoring method based on remote sensing data, comprising:

[0050] The acquisition module is used to acquire the historical fire data of the forest area to be monitored, and the historical fire data comprises fire point data and remote sensing data;

[0051] The combination construction module is used to construct the optimal fire point sensitive waveband group based on the historical fire data: based on the waveband characteristics and the correlation coefficient between the wavebands, the wavebands are combined, and the combination with the maximum OIF index is taken as the optimal fire point sensitive waveband group;

[0052] The representation construction module is used to determine the brightness temperature characteristic representation under the optimal fire point sensitive waveband group;

[0053] The function construction module is used to construct the fire point monitoring index function about the brightness of the micro-light waveband and the vegetation index, and correct the fire point monitoring index function based on the brightness temperature characteristic representation;

[0054] ​The monitoring module is used for inputting real-time remote sensing data of a forest region to be monitored into the corrected fire point monitoring index function to monitor a forest fire point.

[0055] The present application also provides a computer readable medium, which stores a computer program, and the computer program is executed by a processor to realize the forest fire monitoring method based on remote sensing data.

[0056] Compared with the prior art, the present application has the following advantages and beneficial effects:

[0057] The present application provides a forest fire monitoring method, system and medium based on remote sensing data, which improves the fire point identification method on the basis of the existing forest fire monitoring mechanism, combines historical fire data and remote sensing data to construct a fire point monitoring index function, and on the one hand, the fire point monitoring index function considers the micro-light waveband brightness and vegetation index to distinguish forest fires from urban combustion heat sources, and on the other hand, the fire point monitoring index function is corrected based on the brightness temperature feature to distinguish the fire edge position from the urban combustion heat source, so that the identification of the forest fire has sensitivity, stability and accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments, and it should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor. In the drawings:

[0059] Figure 1 It is a flowchart of the forest fire monitoring method based on remote sensing data;

[0060] Figure 2 It is a schematic diagram of the monitoring point;

[0061] Figure 3 It is a schematic diagram of the structure of the forest fire monitoring system based on remote sensing data. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solutions and advantages of the present application more clear, the following will further describe the present application in combination with the embodiments and drawings, and the exemplary embodiments of the present application and the description thereof are only used to explain the present application, and should not be regarded as a limitation on the present application.

[0063] The existing forest fire monitoring mechanism is difficult to separate the forest fire from the urban burning heat source on the remote sensing image by relying on radiation brightness or temperature alone, so that the forest fire monitoring precision is not high and false alarms occur; in view of this, the embodiment provides the following solutions to the above technical problems:

[0064] Embodiment 1

[0065] The embodiment provides a forest fire monitoring method based on remote sensing data, as shown in the following formula: Figure 1

[0066] S1, acquiring historical fire data of a forest region to be monitored, wherein the historical fire data comprises fire point data and remote sensing data;

[0067] S2, constructing an optimal fire point sensitive wave band group based on the historical fire data: performing wave band combination based on wave band characteristics and a correlation coefficient between wave bands, and taking a combination with the maximum OIF index as the optimal fire point sensitive wave band group;

[0068] The step S2 comprises the following sub-steps:

[0069] S21, acquiring the fire point data and extracting non-fire point data corresponding to the fire point data in space from the remote sensing data;

[0070] S22, calculating characteristic values of the fire point data and the non-fire point data in each wave band, wherein the characteristic values comprise a maximum value, a minimum value, an average value and a standard deviation; and finding a wave band with the maximum difference between the characteristic values of the fire point data and the non-fire point data as a basic wave band;

[0071] The basic wave band is acquired by the following method:

[0072] calculating the difference between the maximum value, the minimum value, the average value and the standard deviation of the fire point data and the non-fire point data;

[0073] calculating the arithmetic mean of the difference between the maximum value, the minimum value, the average value and the standard deviation, and screening a wave band with the maximum arithmetic mean as the basic wave band.

[0074] S23, calculating the correlation coefficient between each wave band and the basic wave band, screening a wave band with a correlation coefficient less than a correlation threshold Q A as a first preselected wave band, calculating the correlation coefficient between each first selected wave band, and screening a wave band with a correlation coefficient less than a correlation threshold Q A as a second preselected wave band;

[0075] The correlation coefficient is calculated according to the following formula:

[0076] Q = (δ ij ) 2 ​ / (δ ii +δ jj )

[0077] wherein, (δ ij ) 2 is the covariance of waveband i and waveband j, δ ii is the standard deviation of waveband i, δ jj is the standard deviation of waveband j.

[0078] S24, combining the basic waveband and the second preselected waveband to form a waveband group;

[0079] S25, calculating the OIF index of each waveband group, and taking the waveband group with the maximum OIF index as the optimal fire point sensitive waveband group.

[0080] S3, determining the brightness temperature feature representation under the optimal fire point sensitive waveband group;

[0081] The brightness temperature feature representation comprises a brightness temperature representation of the current position and a brightness temperature difference representation of the current position and the background pixel;

[0082] The brightness temperature representation of the current position is the brightness temperature T e of the current position in the basic waveband;

[0083] The brightness temperature difference representation of the current position and the background pixel is:

[0084]

[0085] wherein, T1 is the brightness temperature of the current position in the first waveband, T2 is the brightness temperature of the current position in the second waveband, T e0 is the average value of the basic waveband background pixel temperature, and the first waveband and the second waveband belong to the optimal fire point sensitive waveband group.

[0086] S4, constructing a fire point monitoring index function about the low-light waveband brightness and the vegetation index, and correcting the fire point monitoring index function based on the brightness temperature feature representation;

[0087] The fire point monitoring index function is:

[0088]

[0089] wherein, k is the fire point monitoring index, NTL G is the normalized low-light waveband brightness, NDVI p is the vegetation index before p time, and a is the fire point monitoring coefficient.

[0090] The correction method of the fire point monitoring index function comprises:

[0091] The fire point monitoring index considering the brightness temperature feature representation is:

[0092]

[0093] Where T eG ΔT represents the normalized brightness temperature of the current position, and ΔT represents the brightness temperature difference between the current position and the background pixel.

[0094] S5 inputs the real-time remote sensing data of the forest area to be monitored into the corrected fire point monitoring index function to monitor forest fire points.

[0095] S5 includes the following sub-steps:

[0096] S51, such as Figure 2 As shown, the map of the forest area to be monitored is divided into an N*N grid. In each square of the grid, the four corners and the center of the square are selected as monitoring points.

[0097] S52, extract the judgment data for each monitoring point from the real-time remote sensing data of the forest area to be monitored: low-light band brightness (NTL) and vegetation index (NDVI) before time p. p And the brightness temperature characteristics of the monitoring points under the optimal fire point sensitive band group;

[0098] S52, input the judgment data of each monitoring point into the fire point monitoring model to calculate the fire point monitoring index k;

[0099] S53, determine whether the fire point monitoring index k exceeds the fire point threshold K. If yes, determine the current monitoring point as a fire point; otherwise, determine the current monitoring point as a safe point.

[0100] S54 distinguishes and displays each fire point on the grid in different ways.

[0101] Different colors are used to represent fire points, or points with equal fire detection indicators are connected together to form isotropic lines. This facilitates the timely location of the fire's center.

[0102] Example 2

[0103] This embodiment provides a forest fire monitoring system based on remote sensing data, used to implement the forest fire monitoring method based on remote sensing data in Embodiment 1, such as... Figure 3 As shown, it includes:

[0104] The data acquisition module is used to acquire historical fire data of the forest area to be monitored, including fire point data and remote sensing data.

[0105] The combination construction module is used to construct the optimal fire point sensitive band group based on historical fire data: bands are combined based on band characteristics and inter-band correlation coefficients, and the combination with the largest OIF index is taken as the optimal fire point sensitive band group.

[0106] a representation constructing module configured to determine a brightness temperature feature representation under the optimal fire point sensitive wave band group;

[0107] a function constructing module configured to construct a fire point monitoring index function about the micro-light wave band brightness and vegetation index, and correct the fire point monitoring index function based on the brightness temperature feature representation;

[0108] a monitoring module configured to input real-time remote sensing data of a forest area to be monitored into the corrected fire point monitoring index function to monitor the forest fire point.

[0109] Embodiment 3

[0110] The embodiment provides a computer readable medium, which stores a computer program, and the computer program is executed by a processor to realize the forest fire monitoring method based on remote sensing data as described in the embodiment 1.

[0111] The above detailed description is further used to explain the purpose, technical scheme and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A forest fire monitoring method based on remote sensing data, characterized by, Comprise: S1, acquire historical fire data of the forest area to be monitored, the historical fire data comprising fire point data and remote sensing data; S2, construct an optimal fire point sensitive band group based on the historical fire data: combine bands based on band characteristics and inter-band correlation coefficients, and take the combination with the largest OIF index as the optimal fire point sensitive band group; specifically comprising: S21, acquiring the fire point data and extracting non-fire point data corresponding to the fire point data in space from the remote sensing data; S22, calculating characteristic values of the fire point data and the non-fire point data in each band, the characteristic values comprising maximum value, minimum value, average value and standard deviation; finding a band with the largest difference between the characteristic values of the fire point data and the non-fire point data as a basic band; the method for acquiring the basic band comprising: calculating the difference between the maximum value, the minimum value, the average value and the standard deviation of the fire point data and the non-fire point data; calculating the arithmetic mean of the difference between the maximum value, the minimum value, the average value and the standard deviation, and screening the band with the largest arithmetic mean as the basic band; S23, calculate the correlation coefficient between each wave band and the basic wave band, and screen out the wave band with the correlation coefficient less than the correlation threshold Q as the first pre-selected wave band, calculate the correlation coefficient between each first selected wave band, and screen out the wave band with the correlation coefficient less than the correlation threshold Q as the second pre-selected wave band A A as the second pre-selected wave band;​ S24, combining the basic band with a second preselected band: combining the basic band and two second preselected bands into a band group; S25, calculating the OIF index of each band group, and taking the band group with the largest OIF index as the optimal fire point sensitive band group; S3, determining a brightness temperature characteristic representation under the optimal fire point sensitive band group; S4, constructing a fire point monitoring index function about the micro-light band brightness and the vegetation index, and correcting the fire point monitoring index function based on the brightness temperature characteristic representation; The brightness temperature characteristic representation comprises a brightness temperature representation of the current position and a brightness temperature difference representation of the current position and a background pixel; The brightness temperature of the current position is expressed as the brightness temperature T of the current position in the basic wave band e ; The brightness temperature difference representation of the current position and the background pixel is: Wherein, T1 is the brightness temperature of the current position in the first wave band, T2 is the brightness temperature of the current position in the second wave band, T e0 is the average value of the basic wave band background pixel temperature, the first wave band and the second wave band belong to the optimal fire point sensitive wave band group S5, inputting real-time remote sensing data of the forest area to be monitored into the corrected fire point monitoring index function for forest fire point monitoring. 2.The forest fire monitoring method based on remote sensing data according to claim 1, characterized in that, The correlation coefficient is calculated according to the following formula Q = (δ ij ) 2 / (δ ii + δ jj ) where δ ij ) 2 is the covariance of band i and band j, δ ii is the standard deviation of band i, δ jj is the standard deviation of band j. 3.The forest fire monitoring method based on remote sensing data according to claim 1, characterized in that, The fire point monitoring index function is: where k is the fire point monitoring index, NTL G is the normalized low-light band brightness, and NDVI p is the vegetation index before p time, and a is the fire point monitoring coefficient.

4. The forest fire monitoring method based on remote sensing data according to claim 3, characterized in that, The correction method of the fire point monitoring index function comprises: The fire point monitoring index considering the brightness temperature characteristic representation is: where T eG is the normalized current location brightness temperature representation, and ΔT is the brightness temperature difference representation of the current location from the background pixel. 5.The forest fire monitoring method based on remote sensing data according to claim 1, characterized in that, S5 comprises the following sub-steps: S51, dividing a map of the forest area to be monitored into N*N grids, and selecting the positions of the four corners and the center of each grid as monitoring points in each grid; S52, extract the judgment data for each monitoring point from the real-time remote sensing data of the forest area to be monitored: low-light band brightness (NTL) and vegetation index (NDVI) before time p. p And the brightness temperature characteristics of the monitoring points under the optimal fire point sensitive band group; S52, inputting the determination data of each monitoring point into the fire point monitoring model to calculate the fire point monitoring index k; S53, judging whether the fire point monitoring index k exceeds a fire point threshold K, if yes, determining that the current monitoring point is a fire point; otherwise, determining that the current monitoring point is a safe point; S54, distinguishing and displaying each fire point on the grid in different forms.

6. A forest fire monitoring system based on remote sensing data, characterized in that, The method for implementing the forest fire monitoring method based on remote sensing data according to any one of claims 1-5, comprising: an acquisition module configured to acquire historical fire data of the forest area to be monitored, the historical fire data comprising fire point data and remote sensing data; a combination construction module configured to construct an optimal fire point sensitive band group based on the historical fire data: combine bands based on band characteristics and inter-band correlation coefficients, and take the combination with the largest OIF index as the optimal fire point sensitive band group; a representation constructing module configured to determine a brightness temperature feature representation under the optimal fire point sensitive wave band group; a function constructing module configured to construct a fire point monitoring index function about the micro-light wave band brightness and vegetation index, and correct the fire point monitoring index function based on the brightness temperature feature representation; a monitoring module configured to input real-time remote sensing data of a forest region to be monitored into the corrected fire point monitoring index function to monitor a forest fire point.

7. A computer readable medium having stored thereon a computer program, characterized in that The computer program is executed by a processor to implement the forest fire monitoring method based on remote sensing data according to any one of claims 1-5.

Images