An intelligent monitoring method for forest fires based on remote sensing images

Abstract

The present application belongs to the field of image data processing technology, and provides a remote sensing image-based intelligent forest fire monitoring method, including determining a suspected smoke area in a remote sensing image; determining a smoke possibility index of the suspected smoke area according to the distribution of colors and the brightness of pixels in the suspected smoke area; analyzing the position offset of the suspected smoke area in two adjacent remote sensing image frames to obtain a smoke offset difference value of the suspected smoke area; analyzing the difference characteristics and distance characteristics of smoke possibility indicators of adjacent suspected smoke areas, combining the smoke offset difference value, obtaining the possibility of smoke divergence position, and then determining the smoke divergence area; and monitoring forest fires according to the smoke divergence area. The present invention distinguishes smoke areas in suspected smoke areas through the internal color distribution characteristics of smoke and the aggregation and expansion characteristics of smoke, thereby improving the accuracy of forest fire judgment.

Description

Technical Field

[0001] The present application relates to the technical field of image data processing, and in particular to an intelligent forest fire monitoring method based on remote sensing images. Background Art

[0002] Forest fires are fires that occur in vast forests due to natural or human factors. Forest fires are characterized by strong suddenness, great destructiveness, and high danger. They are one of the most frequent, most difficult to deal with, and most serious natural disasters in the world. They are the biggest threat to the achievements of ecological civilization construction and the safety of forest resources, and may even cause ecological disasters and social crises. Therefore, high-precision, full-coverage, and timely monitoring and early warning of forest fires can prevent and contain forest fire accidents.

[0003] Remote sensing images of forest areas can be collected by satellites, and then through the analysis of remote sensing images, comprehensive and real-time monitoring of forest fires can be achieved. Remote sensing images can capture the location of the fire point and the spread of the fire, etc., without entering the fire scene, which can improve monitoring efficiency, detect fires in time, reduce losses and reduce risks to personnel safety.

[0004] At present, the analysis of remote sensing images of forest fires mainly relies on color changes. When a fire occurs in a forest area, a large amount of smoke usually appears first, and the smoke is generally light-colored in remote sensing images; however, when sunlight shines directly on vegetation, the remote sensing images obtained will also have light-colored parts. The two cannot be distinguished in remote sensing images by color alone, which leads to poor reliability in judging forest fires. Summary of the invention

[0005] In order to solve the above technical problems, the present application provides a forest fire intelligent monitoring method based on remote sensing images.

[0006] According to a remote sensing image-based intelligent forest fire monitoring method provided in this application, the method includes:

[0007] Setting up satellite data receiving stations in forests to obtain remote sensing images of forest areas;

[0008] Acquire a remote sensing image frame in the remote sensing image, and determine a suspected smoke area in the remote sensing image frame;

[0009] Determining a smoke possibility index of the suspected smoke area according to the distribution of colors and the brightness of pixels in the suspected smoke area;

[0010] Analyze the position offset of the suspected smoke area in two adjacent remote sensing image frames to obtain a smoke offset difference value of the suspected smoke area;

[0011] Determine the suspected smoke area adjacent to the suspected smoke area, analyze the smoke possibility index difference characteristics and distance characteristics of the adjacent suspected smoke area, and obtain the possibility that the suspected smoke area is the smoke divergence location in combination with the smoke offset difference value;

[0012] Determining a smoke dispersion area according to the possibility of the smoke dispersion position;

[0013] Forest fire monitoring is carried out based on the smoke dispersion area.

[0014] In some embodiments of the present invention, obtaining a remote sensing image frame in the remote sensing image, and determining a suspected smoke area in the remote sensing image frame includes:

[0015] Acquire a remote sensing image frame from the remote sensing image, and convert the remote sensing image frame into an HSV image;

[0016] Pixels whose brightness and saturation values ​​in the HSV image meet a preset range are screened to obtain a connected domain of suspected smoke pixels, and the connected domain of suspected smoke pixels is determined to be a suspected smoke area.

[0017] In some embodiments of the present invention, after determining the suspected smoke area, the method further includes:

[0018] The HSV image of the suspected smoke area is converted into an RGB image, and the RGB image is gray-scaled to obtain a gray-scale image of the suspected smoke area in the remote sensing image frame.

[0019] In some embodiments of the present invention, determining the smoke possibility index of the suspected smoke area according to the distribution of colors and the brightness of pixels in the suspected smoke area includes:

[0020] Analyze the characteristics of the V channel and the S channel of the pixel points at the suspected smoke area position in the HSV image to obtain the color distribution;

[0021] Analyze the grayscale value fluctuation of the pixel points in the grayscale image of the suspected smoke area to obtain the pixel point brightness;

[0022] A smoke possibility index of the suspected smoke area is determined according to the color distribution and the pixel brightness.

[0023] In some embodiments of the present invention, the calculation formula of the smoke possibility index is:

[0024]

[0025] Among them, A v represents the smoke possibility index of the vth suspected smoke area, Nv Indicates the number of pixels in the vth suspected smoke area, V v,i and S v,i represents the value of the V channel and S channel of the ith pixel in the vth suspected smoke area, G v,i represents the gray value of the i-th pixel in the v-th suspected smoke area, Represents the mean grayscale value of all pixels in the vth suspected smoke area.

[0026] In some embodiments of the present invention, the difference characteristics between the grayscale values ​​of the pixels at corresponding positions in the suspected smoke area in two adjacent remote sensing image frames are analyzed to obtain the smoke offset difference value of the suspected smoke area.

[0027] In some embodiments of the present invention, the calculation formula of the smoke offset difference value is:

[0028]

[0029] Among them, D v Indicates the smoke offset difference value of the vth suspected smoke area, N v represents the number of pixels in the vth suspected smoke area, G v,i represents the gray value of the i-th pixel in the v-th suspected smoke area of ​​the current frame, G v ′ ,i Indicates the grayscale value of the pixel in the previous frame that is at the same position as the i-th pixel in the current frame.

[0030] In some embodiments of the present invention, analyzing the smoke possibility index difference characteristics and distance characteristics adjacent to the suspected smoke area includes:

[0031] Obtaining a difference between the smoke possibility index of the suspected smoke area and the smoke possibility index of an adjacent suspected smoke area;

[0032] Comparing the difference of the smoke possibility index with the smoke possibility index of the suspected smoke area to obtain the difference feature of the smoke possibility index;

[0033] The distance between the center point of the suspected smoke area and the center point of its adjacent suspected smoke area is analyzed to obtain a distance feature.

[0034] In some embodiments of the present invention, the calculation formula for the probability of the smoke emission position is:

[0035]

[0036] Among them, P vrepresents the probability of the smoke divergence position of the vth suspected smoke area, N represents the number of neighboring suspected smoke areas of the vth suspected smoke area, and A v represents the smoke possibility index determined in the vth suspected smoke area, A v,i represents the smoke probability index determined in the ith neighboring suspected smoke area of ​​the vth suspected smoke area, L v,i represents the distance between the ith adjacent suspected smoke area of ​​the vth suspected smoke area and the vth suspected smoke area, D v represents the smoke offset difference value of the vth suspected smoke area, softmax represents the normalization function, and e represents the exponential with the natural constant e as the base.

[0037] In some embodiments of the present invention, determining the smoke emitting area according to the possibility of the smoke emitting position includes:

[0038] Set the probability threshold of the smoke emission location;

[0039] Comparing the probability of the smoke emanation location with a probability threshold of the smoke emanation location;

[0040] If the probability of the smoke diverging position is greater than or equal to the probability threshold of the smoke diverging position, determining the corresponding suspected smoke area as the smoke diverging area;

[0041] If the probability of the smoke divergence position is less than the probability threshold of the smoke divergence position, it is determined that the corresponding suspected smoke area is a vegetation area.

[0042] It can be seen from the above embodiments that the intelligent forest fire monitoring method based on remote sensing images provided in the embodiments of the present application has the following beneficial effects:

[0043] The present invention obtains remote sensing images of forest areas by setting a satellite data receiving station in the forest; obtains remote sensing image frames in the remote sensing images, determines suspected smoke areas in the remote sensing image frames; determines smoke possibility indicators of the suspected smoke areas according to the distribution of colors and the brightness of pixels in the suspected smoke areas; analyzes the position offset of the suspected smoke areas in two adjacent remote sensing image frames to obtain smoke offset difference values ​​of the suspected smoke areas; determines adjacent suspected smoke areas of the suspected smoke areas, analyzes the smoke possibility indicator difference characteristics and distance characteristics of the adjacent suspected smoke areas, and obtains the possibility that the suspected smoke areas are smoke divergence locations in combination with the smoke offset difference values; determines smoke divergence areas according to the possibility of the smoke divergence locations; and monitors forest fires according to the smoke divergence areas. The present invention identifies smoke areas in suspected smoke areas by the internal color distribution characteristics of smoke and the aggregation and expansion characteristics of smoke. The accuracy of determining smoke areas is improved, and at the same time, the system is reported to carry out timely and effective early warning and response before the fire spreads.

[0044] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 A basic flow chart of a remote sensing image-based intelligent forest fire monitoring method provided in an embodiment of the present application;

[0047] Figure 2 A schematic diagram of a basic flow chart of a method for determining a suspected smoke area provided in an embodiment of the present application;

[0048] Figure 3 A schematic diagram of the basic flow of a method for determining a smoke possibility index of a suspected smoke area provided in an embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a forest fire intelligent monitoring method based on remote sensing images proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0050] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0051] The open fire warning method based on multi-sensor data provided by this embodiment will be described in detail below with reference to the accompanying drawings.

[0052] Figure 1 A basic flow chart of a remote sensing image-based intelligent forest fire monitoring method provided in an embodiment of the present application; Figure 2 A schematic diagram of a basic flow chart of a method for determining a suspected smoke area provided in an embodiment of the present application; Figure 3 A schematic diagram of the basic flow of a method for determining a smoke possibility index of a suspected smoke area provided in an embodiment of the present application.

[0053] When a fire occurs in a forest, there are large areas of light-colored smoke in the remote sensing image of the forest, which are fire areas, and there are also light-colored vegetation areas due to illumination. The two cannot be accurately distinguished by color alone, making it difficult to accurately determine the fire area. Therefore, the purpose of the present invention is to distinguish light-colored smoke caused by fire and light-colored vegetation areas due to illumination in remote sensing images of the forest by using the distribution and diffusion characteristics of smoke.

[0054] like Figure 1 As shown, this embodiment provides a forest fire intelligent monitoring method based on remote sensing images, which specifically includes the following steps:

[0055] S100: Set up a satellite data receiving station in the forest to obtain remote sensing images of the forest area.

[0056] The embodiment of the present application needs to obtain remote sensing images of the forest. A special satellite data receiving station is set up in the forest to receive remote sensing image data from satellites. The satellite data receiving station can obtain real-time remote sensing images of the forest through the data stream transmitted by the satellite.

[0057] Since the main purpose of this embodiment is to distinguish forest fire smoke areas and vegetation areas caused by illumination, and both areas appear light in remote sensing images, it is necessary to first obtain the suspected smoke area in the remote sensing image, further determine the possibility that the suspected smoke area is a smoke area through the distribution characteristics of forest fire smoke, and then identify the location of forest fire smoke in the suspected smoke area in combination with the diffusion characteristics of forest fire smoke. The specific implementation method is steps S200 to S600.

[0058] S200: Acquire a remote sensing image frame in the remote sensing image, and determine a suspected smoke area in the remote sensing image frame.

[0059] In remote sensing images, since forest fire smoke areas and vegetation areas caused by light are light-colored, it is first necessary to obtain remote sensing image frames in the remote sensing images and determine the suspected smoke areas in the remote sensing image frames.

[0060] like Figure 2 As shown, a remote sensing image frame is obtained from the remote sensing image, and a suspected smoke area is determined in the remote sensing image frame, including:

[0061] S201: Acquire a remote sensing image frame from a remote sensing image, and convert the remote sensing image frame into an HSV image.

[0062] The remote sensing image frame of the acquired remote sensing image is an RGB image. Since light colors have a more accurate threshold range in the HSV image, the RGB image of the remote sensing image frame is first converted into an HSV image, where H refers to hue, S is saturation, and V is brightness.

[0063] S202: Filtering pixels whose brightness and saturation values ​​satisfy a preset range in the HSV image, obtaining a connected domain of suspected smoke pixels, and determining the connected domain of suspected smoke pixels as a suspected smoke area.

[0064] First, determine the thresholds of each HSV channel corresponding to light colors. For the suspected smoke area in the remote sensing image frame, the suspected smoke area has different brightness, so the range of HSV values ​​can be set. Since the brightness is from light to dark, the value of the brightness (V) channel will gradually increase, so V = [0.5, 1] ​​can be selected; and since the saturation of light colors is relatively high, the saturation (S) value is usually close to or equal to 1, so S = [0.9, 1] can be selected; it does not involve the change of the basic hue of the color, so H can take any value.

[0065] After determining the thresholds of each HSV channel corresponding to the light color, the pixels whose brightness and saturation values ​​in the HSV image meet the preset range are screened to obtain the connected domain of the suspected smoke pixels, and the connected domain of the suspected smoke pixels is determined to be the suspected smoke area.

[0066] After the suspected smoke area is identified, it also includes:

[0067] S203: Convert the HSV image of the suspected smoke area into an RGB image, perform grayscale processing on the RGB image, and obtain a grayscale image of the suspected smoke area in the remote sensing image frame.

[0068] The image of the suspected smoke area obtained by step S202 is an HSV image. In order to facilitate subsequent analysis, after obtaining the suspected smoke area, it is necessary to convert the HSV image of the suspected smoke area into an RGB image, and perform grayscale processing on the RGB image to obtain a grayscale image of the suspected smoke area corresponding to the remote sensing image frame.

[0069] S300: Determine a smoke possibility index of the suspected smoke area according to the distribution of colors and the brightness of pixels in the suspected smoke area.

[0070] The smoke from forest fires is floating, and the light color of the smoke is affected by factors such as smoke concentration and thickness. When the smoke concentration is enough to cover the color of vegetation, the smoke appears as a distinct light color in the remote sensing image; when the smoke thickness is thin or the concentration is low, the smoke may appear brighter in the remote sensing image; therefore, the overall gray value data of the forest fire smoke area fluctuates greatly.

[0071] Vegetation leaves usually have a high reflectivity. When sunlight shines directly on the surface of vegetation leaves, the leaves will reflect a large amount of light. In addition, the density of vegetation in the forest is relatively high, and the reflected light is also relatively large, causing the vegetation area to appear as a darker suspected smoke area in the remote sensing image. Therefore, the values ​​of the S channel and the V channel in the pixels belonging to the surface of vegetation leaves are closer to 1. At the same time, the sunlight is far away from the vegetation, and horizontal reflections will be generated when the light passes through the vegetation leaves. The vegetation leaves are complex and intertwined, and these reflected lights will interweave in space, resulting in the overall uniformity of the color presented in the remote sensing image, so the overall gray value of the area is relatively uniform.

[0072] Based on the above analysis, if Figure 3 As shown, based on the distribution of colors in the suspected smoke area and the brightness of the pixels, the smoke possibility index of the suspected smoke area is determined, including:

[0073] S301: Analyze the characteristics of the V channel and the S channel of the pixel points at the suspected smoke area position in the HSV image to obtain the color distribution.

[0074] Since the values ​​of the V and S channels of the pixels belonging to the surface of vegetation leaves are closer to 1 than those of the pixels belonging to smoke, the characteristics of the V and S channels of the pixels at the suspected smoke area in the HSV image are analyzed, that is, the V channel value V of the pixels is compared.v,i and the value S of the S channel v,i The difference between 1 and v,i -1) and (S v,i -1); the value of the V channel of the pixel point V v,i and the value S of the S channel v,i The larger the difference from 1, the higher the value of the V channel of the pixel. v,i and the value S of the S channel v,i The further away from 1, the darker the pixel is, which means the pixel is more likely to be smoke from a forest fire. v,i and the value S of the S channel v,i The smaller the difference from 1, the higher the value of the V channel of the pixel. v,i and the value S of the S channel v,i The closer it is to 1, the brighter the pixel is, and the more likely it is that the pixel is a light-colored vegetation area affected by light. The difference between the V channel and S channel values ​​of the ith pixel in the vth suspected smoke area and 1 is summed, and the color distribution of the pixels in the suspected smoke area is finally obtained as [(V v,i -1)+(S v,i -1)].

[0075] S302: Analyze the grayscale value fluctuation of the pixel points in the grayscale image of the suspected smoke area to obtain the brightness of the pixel points.

[0076] Since the overall grayscale value data of the smoke area of ​​the forest fire fluctuates greatly, and the overall grayscale value of the light-colored vegetation area affected by light is relatively uniform, the grayscale value fluctuation of the pixel point in the grayscale image of the suspected smoke area is analyzed, that is, the grayscale value G of the pixel point is calculated. v,i The mean gray value of all pixels in the suspected smoke area where the pixel is located The difference between the two is the pixel brightness of the pixel in the suspected smoke area. The larger the value, the gray value G of the pixel point v,i The mean gray value of all pixels in the suspected smoke area where the pixel is located The larger the difference, the more obvious the fluctuation of the gray value of the pixel is, which means that the pixel is more likely to be smoke generated by forest fire; conversely, the smaller the value, the gray value of the pixel is. v,i The mean gray value of all pixels in the suspected smoke area where the pixel is located The smaller the difference, the less obvious the fluctuation of the gray value of the pixel is, which means that the pixel is more likely to be a light-colored vegetation area affected by light.

[0077] S303: Determine a smoke possibility index of the suspected smoke area according to the color distribution and the pixel brightness.

[0078] The color distribution of the pixels in the suspected smoke area and the brightness of the pixels together indicate the possibility that the suspected smoke area is smoke. Therefore, the color distribution of the pixels in the suspected smoke area obtained in steps S301 and S302 [(V v,i -1)+(S v,i -1)] and pixel brightness Multiplying by , the probability index of determining that the pixel point in the suspected smoke area is the pixel point corresponding to the smoke is Then, all the pixels in the vth suspected smoke area are summed and averaged, and the final color distribution is:

[0079] That is, the calculation formula of the smoke possibility index is:

[0080]

[0081] Among them, A v represents the smoke possibility index of the vth suspected smoke area, N v Indicates the number of pixels in the vth suspected smoke area, V v,i and S v,i represents the value of the V channel and S channel of the ith pixel in the vth suspected smoke area, G v,i represents the gray value of the i-th pixel in the v-th suspected smoke area, Represents the mean grayscale value of all pixels in the vth suspected smoke area.

[0082] The larger the smoke possibility index value of the suspected smoke area, the darker the color of the vth suspected smoke area and the greater the fluctuation of the grayscale value of the pixel, which means that the vth suspected smoke area is more likely to be smoke generated by forest fire; conversely, the smaller the value, the brighter the color of the vth suspected smoke area and the smaller the fluctuation of the grayscale value of the pixel, which means that the vth suspected smoke area is more likely to be a light-colored vegetation area affected by light.

[0083] S400: Analyze the position offset of the suspected smoke area in two adjacent remote sensing image frames to obtain the smoke offset difference value of the suspected smoke area.

[0084] The position offset of the suspected smoke area in two adjacent remote sensing image frames is analyzed, and the difference characteristics between the grayscale values ​​of the pixels at corresponding positions in the suspected smoke area in the two adjacent remote sensing image frames are further analyzed to obtain the smoke offset difference value of the suspected smoke area.

[0085] Since smoke will spread due to wind or other factors, the position of smoke will change in different remote sensing image frames. Since the position of smoke has changed, the position of the suspected smoke area determined in the current frame is different from that in the previous frame. Therefore, the smoke area in the current frame is not the smoke area in the previous frame (i.e., the suspected smoke area in the remote sensing image). Therefore, it is more likely that the grayscale value of the smoke area in the current frame will change compared with the same position in the previous frame. The position of vegetation is generally fixed, so the position of vegetation in different frames generally does not change. Therefore, it is less likely that the grayscale value of the light-colored vegetation area in the current frame will change compared with the same position in the previous frame.

[0086] Based on the above analysis, it is necessary to determine the position offset of the suspected smoke area in the adjacent frames. For any suspected smoke area in the current frame, determine the grayscale value of the pixel at the same position in the current frame and the previous frame, and calculate the grayscale value difference between the pixels at the same position in the two frames to determine the smoke offset difference value.

[0087] The calculation formula for the smoke offset difference value is:

[0088]

[0089] Among them, D v Indicates the smoke offset difference value of the vth suspected smoke area, N v represents the number of pixels in the vth suspected smoke area, G v,i represents the gray value of the i-th pixel in the v-th suspected smoke area of ​​the current frame, G v ′ ,i Indicates the grayscale value of the pixel in the previous frame that is at the same position as the i-th pixel in the current frame.

[0090] |G v,i -G v ′ ,i The larger the | value is, the larger the difference between the grayscale value of the ith pixel in the vth suspected smoke area in the previous frame and the grayscale value of the pixel at the same position in the previous frame is, which means that the possibility that the vth suspected smoke area is smoke produced by a forest fire is greater; conversely, the smaller the value is, the smaller the difference between the grayscale value of the ith pixel in the vth suspected smoke area in the previous frame and the grayscale value of the pixel at the same position in the previous frame is, which means that the possibility that the vth suspected smoke area is a light-colored vegetation area affected by light is greater.

[0091] S500: Analyze the smoke possibility index difference characteristics and distance characteristics of the adjacent suspected smoke area, and combine the smoke offset difference value to obtain the possibility that the suspected smoke area is the smoke divergence location.

[0092] Since smoke will spread due to wind or other factors, the position of smoke will change in different remote sensing image frames. At the same time, when smoke from the same fire source begins to spread, the distance between smoke areas will not be very far, and the similarity between smoke areas is also relatively high. Therefore, this embodiment can combine the distribution characteristics and distance characteristics between suspected smoke areas in adjacent suspected smoke areas, as well as the position changes of suspected smoke areas in adjacent frames, to determine whether a known suspected smoke area is the divergence position of smoke.

[0093] Based on the above analysis, the difference characteristics and distance characteristics of the smoke possibility index of the neighboring suspected smoke area are analyzed, and the smoke offset difference value is combined to obtain the possibility that the suspected smoke area is the smoke divergence position. Further, for any suspected smoke area, the N (in the present invention, N=3) suspected smoke areas closest to the current suspected smoke area are determined, and the N suspected smoke areas are determined as the neighboring suspected smoke areas of the current suspected smoke area. Obtain the difference between the smoke possibility index of the suspected smoke area and the smoke possibility index of the neighboring suspected smoke area; compare the difference in the smoke possibility index with the smoke possibility index of the suspected smoke area to obtain the difference characteristics of the smoke possibility index; analyze the distance between the center point of the suspected smoke area and the center point of the neighboring suspected smoke area to obtain the distance characteristics. Among them, the center point of the suspected smoke area is, determine the circumscribed rectangle of all suspected smoke areas, and determine the center point of the outer rectangle as the center point of the suspected smoke area. According to the difference characteristics and distance characteristics of the smoke possibility index, combined with the smoke offset difference value, the possibility of the suspected smoke area being the smoke divergence position is obtained. The calculation formula for the possibility of the smoke divergence position is:

[0094]

[0095] Among them, P v represents the probability of the smoke divergence position of the vth suspected smoke area, N represents the number of neighboring suspected smoke areas of the vth suspected smoke area, and A v represents the smoke possibility index determined in the vth suspected smoke area, A v,i represents the smoke probability index determined in the ith neighboring suspected smoke area of ​​the vth suspected smoke area, L v,i represents the distance between the ith adjacent suspected smoke area of ​​the vth suspected smoke area and the vth suspected smoke area, D v It represents the smoke offset difference value of the vth suspected smoke area, and e represents the exponent with the natural constant e as the base.

[0096] |A v -A v,i| represents the difference in smoke possibility index between the adjacent suspected smoke area and the current suspected smoke area. The smaller the value, the more similar the adjacent suspected smoke area is to the current suspected smoke area, and the greater the possibility that the adjacent suspected smoke area and the current suspected smoke area are the same suspected smoke area. v Indicates the smoke possibility index of the current suspected smoke area. The larger the value, the greater the possibility that the current suspected smoke area is a smoke dispersion area; It means that on the basis of a larger smoke possibility index, the smaller the difference between the adjacent suspected smoke area and the current suspected smoke area, the smaller the value, indicating that the current suspected smoke area is more likely to be a smoke area; conversely, it means that on the basis of a smaller smoke possibility index, the greater the similarity between the adjacent suspected smoke area and the current suspected smoke area, the larger the value, indicating that the current suspected smoke area is more likely to be a vegetation area. v,i Indicates the distance between the adjacent suspected smoke area and the current suspected smoke area. The larger the value, the farther the current suspected smoke area is from the adjacent suspected smoke area, indicating that the possibility that the suspected smoke area is a smoke divergence area is smaller. Conversely, the smaller the value, the closer the current suspected smoke area is from the adjacent suspected smoke area, indicating that the possibility that the suspected smoke area is a smoke divergence area is greater. v Indicates the smoke offset difference value of the current suspected smoke area. The larger the value, the corresponding The smaller the value, the greater the deviation of the suspected smoke area, indicating that the current suspected smoke area is more likely to belong to the smoke divergence area. From the above analysis, it can be seen that the probability of each part and the smoke divergence position is inversely proportional. Therefore, after the data is normalized by softmax, it is processed by 1 minus.

[0097] S600: Determine a smoke dispersion area according to the possibility of the smoke dispersion position.

[0098] According to the possibility of the smoke divergence position, the smoke divergence area is determined, and further: a possibility threshold of the smoke divergence position is set. The possibility threshold of the smoke divergence position in this embodiment can be set to 0.7; the possibility of the smoke divergence position is compared with the possibility threshold of the smoke divergence position; if the possibility of the smoke divergence position is greater than or equal to the possibility threshold of the smoke divergence position, that is, P v ≥0.7, the corresponding suspected smoke area is determined to be a smoke divergence area; if the probability of the smoke divergence location is less than the probability threshold of the smoke divergence location, that is, P v <0.7, the corresponding suspected smoke area is determined to be a vegetation area.

[0099] S700: Monitor forest fires based on smoke emission areas.

[0100] Obtain an electronic map of the forest location, determine the location of the smoke dispersion area in the forest, use a marking tool or icon to mark the location of the smoke point on the electronic map, save the marked electronic map file, and upload the electronic map file to the monitoring system for relevant personnel to view.

[0101] Through the method provided in this embodiment, the location of forest fire smoke points can be visualized, and monitoring personnel can be helped to understand the fire situation in real time so as to take appropriate measures for effective response and handling.

[0102] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0103] It should be noted that, unless otherwise specified and limited, terms such as "include", "comprise" or any other variations thereof are intended to cover non-exclusive inclusion, so that a circuit structure, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such article or device. In the absence of more restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the article or device including the element.

[0104] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention invented herein. This application is intended to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary technical means in the art not invented by the present invention.

[0105] It should be understood that the present invention is not limited to the above description, and various modifications and changes may be made without departing from the scope thereof, and various modifications and changes made without departing from the scope thereof are within the protection scope of the present invention.

Claims

1. A forest fire intelligent monitoring method based on remote sensing images, characterized in that: The method comprises: Setting up satellite data receiving stations in forests to obtain remote sensing images of forest areas; Acquire a remote sensing image frame in the remote sensing image, and determine a suspected smoke area in the remote sensing image frame; Determining a smoke possibility index of the suspected smoke area according to the distribution of colors and the brightness of pixels in the suspected smoke area; Analyze the position offset of the suspected smoke area in two adjacent remote sensing image frames to obtain a smoke offset difference value of the suspected smoke area; Determine the suspected smoke area adjacent to the suspected smoke area, analyze the smoke possibility index difference characteristics and distance characteristics of the adjacent suspected smoke area, and obtain the possibility that the suspected smoke area is the smoke divergence location in combination with the smoke offset difference value; Determining a smoke dispersion area according to the possibility of the smoke dispersion position; Conduct forest fire monitoring based on the smoke dispersion area; Acquiring a remote sensing image frame from the remote sensing image, and determining a suspected smoke area in the remote sensing image frame, including: Acquire a remote sensing image frame from the remote sensing image, and convert the remote sensing image frame into an HSV image; Filtering pixels whose brightness and saturation values ​​in the HSV image meet a preset range, obtaining a connected domain of suspected smoke pixels, and determining the connected domain of suspected smoke pixels as a suspected smoke area; Determining a smoke possibility index of the suspected smoke area according to the distribution of colors and the brightness of pixels in the suspected smoke area includes: Analyze the characteristics of the V channel and the S channel of the pixel points at the suspected smoke area position in the HSV image to obtain the color distribution; Analyze the grayscale value fluctuation of the pixel points in the grayscale image of the suspected smoke area to obtain the pixel point brightness; Determining a smoke possibility index of the suspected smoke area according to the color distribution and the pixel brightness; The calculation formula of the smoke possibility index is: in, Indicates Smoke probability indicators for suspected smoke areas, Indicates The number of pixels in the suspected smoke area, and Indicates Among the suspected smoke areas The values ​​of the V channel and S channel of each pixel, Indicates Among the suspected smoke areas The gray value of a pixel, Indicates The mean gray value of all pixels in the suspected smoke area; Analyze the difference characteristics and distance characteristics of the smoke possibility indicators adjacent to the suspected smoke area, including: Obtaining a difference between the smoke possibility index of the suspected smoke area and the smoke possibility index of an adjacent suspected smoke area; Comparing the difference of the smoke possibility index with the smoke possibility index of the suspected smoke area to obtain the difference feature of the smoke possibility index; Analyze the distance between the center point of the suspected smoke area and the center points of the adjacent suspected smoke areas to obtain a distance feature; The calculation formula for the probability of the smoke dispersion position is: in, Indicates The probability of smoke emission from a suspected smoke area. Indicates The number of neighboring suspected smoke areas of a suspected smoke area, Indicates The smoke probability index determined in the suspected smoke area, Indicates The suspected smoke area The smoke probability index determined in the adjacent suspected smoke area, Indicates The suspected smoke area The adjacent suspected smoke area and The distance between the suspected smoke areas, Indicates The smoke offset difference value of the suspected smoke area, represents the normalization function, Indicated by natural constant The exponent of the base.

2. The method for intelligent monitoring of forest fires based on remote sensing images according to claim 1 is characterized in that: After the suspected smoke area is identified, it also includes: The HSV image of the suspected smoke area is converted into an RGB image, and the RGB image is gray-scaled to obtain a gray-scale image of the suspected smoke area in the remote sensing image frame.

3. The method for intelligent monitoring of forest fires based on remote sensing images according to claim 2 is characterized in that: The difference characteristics between the grayscale values ​​of the pixels at corresponding positions in the suspected smoke area in two adjacent remote sensing image frames are analyzed to obtain the smoke offset difference value of the suspected smoke area.

4. The method for intelligent monitoring of forest fires based on remote sensing images according to claim 3 is characterized in that: The calculation formula of the smoke offset difference value is: in, Indicates The smoke offset difference value of the suspected smoke area, Indicates The number of pixels in the suspected smoke area, Indicates the current frame Among the suspected smoke areas The gray value of a pixel, Indicates the previous frame and the current frame. The grayscale value of pixels at the same position of pixels.

5. The method for intelligent monitoring of forest fires based on remote sensing images according to claim 1 is characterized in that: According to the possibility of the smoke dispersion position, the smoke dispersion area is determined, including: Set the probability threshold of the smoke emission location; Comparing the probability of the smoke emanation location with a probability threshold of the smoke emanation location; If the probability of the smoke diverging position is greater than or equal to the probability threshold of the smoke diverging position, determining the corresponding suspected smoke area as the smoke diverging area; If the probability of the smoke divergence position is less than the probability threshold of the smoke divergence position, it is determined that the corresponding suspected smoke area is a vegetation area.

Images