A Visual Analysis Method and System for LiDAR Fire Monitoring in Forest Protection

Abstract

This invention discloses a laser radar method and system for visual analysis of fire monitoring in forest protection, relating to the field of visual analysis technology for fire monitoring. The method includes the following steps: acquiring a dynamic point cloud region based on a continuous two-dimensional point cloud image; acquiring an image of the dynamic point cloud region in the projection direction of the two-dimensional point cloud image and marking it as a detection image; acquiring a first number of historical flame images; performing grayscale processing on the historical flame images to obtain historical flame grayscale images; acquiring a flame grayscale threshold range based on the historical flame grayscale images; and determining whether the dynamic point cloud region is an abnormal region based on the flame grayscale threshold range and the detection image. If so, a signal indicating the occurrence of a flame is issued. This invention addresses the problem in existing laser radar fire monitoring technology where environmental changes lead to low accuracy in determining whether fire has occurred based on changes in the laser radar point cloud image.

Description

Technical Field

[0001] This invention relates to the field of visual analysis technology for fire monitoring, specifically to a laser radar method and system for visual analysis of fire monitoring used in forest protection. Background Technology

[0002] Forest fires are one of the world's eight major natural disasters, characterized by their suddenness, destructiveness, high danger, and difficulty in handling and fighting them. They seriously endanger people's lives and property and the safety of forest resources. Traditional forest fire prevention and monitoring methods mainly rely on manual ground patrols and high-point patrols from lookout towers, which have problems such as limited patrol range, low efficiency, and being limited by environmental conditions.

[0003] With the development of technology, point cloud maps of forests can be obtained based on lidar. When a fire occurs, the obtained point cloud map will change, so the changes in the point cloud map can be monitored in real time to determine whether a fire has occurred. However, there are many factors that cause changes in forest point cloud maps, such as animal movement. This makes it not entirely accurate to determine whether a fire has occurred in a forest. In other words, the existing lidar fire monitoring technology has low accuracy in determining whether a fire has occurred based on changes in lidar point cloud maps due to the influence of environmental changes. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in the prior art. It obtains a dynamic point cloud region based on a continuous two-dimensional point cloud map, acquires an image of the dynamic point cloud region in the projection direction of the two-dimensional point cloud map, marks it as a detection map, acquires a first number of historical flame images, performs grayscale processing on the historical flame images to obtain historical flame grayscale images, obtains a flame grayscale threshold range based on the historical flame grayscale images, and determines whether the dynamic point cloud region is an abnormal region based on the flame grayscale threshold range and the detection map. If so, it issues a flame generation signal. This solves the problem in the existing lidar smoke and fire monitoring technology where the accuracy of judging whether smoke and fire have occurred based on changes in lidar point cloud maps is low due to environmental changes.

[0005] To achieve the above objectives, this application provides a laser radar method for visual analysis of smoke and fire monitoring in forest protection, comprising the following steps:

[0006] Forest point cloud images are continuously acquired using lidar, and a two-dimensional projection of the forest point cloud images is obtained and marked as a two-dimensional point cloud image.

[0007] Dynamic point cloud regions are obtained based on continuous two-dimensional point cloud maps;

[0008] The image of the dynamic point cloud region is obtained in the projection direction of the two-dimensional point cloud map and marked as the detection map;

[0009] Obtain the first number of historical flame images, and perform grayscale processing on the historical flame images to obtain historical flame grayscale images;

[0010] Obtain the flame grayscale threshold range based on historical flame grayscale images;

[0011] Based on the flame grayscale threshold range and the detection map, it is determined whether the dynamic point cloud region is an abnormal region. If so, a signal to generate a flame is issued.

[0012] Furthermore, obtaining a dynamic point cloud region based on a continuous two-dimensional point cloud map includes the following sub-steps:

[0013] The real-time acquired 2D point cloud map is marked as the first point cloud map, and the 2D point cloud map acquired in the previous frame is marked as the second point cloud map;

[0014] Establish a Cartesian coordinate system, labeled as the cloud point coordinate system. Place the first cloud map in the first quadrant and place the second cloud map in the same position as the first cloud map.

[0015] Mark each cloud point in the first cloud map as the first cloud point;

[0016] Mark each cloud point in the second point cloud map as a second cloud point;

[0017] In the cloud point coordinate system, the second cloud point that is closest to the first cloud point is marked as the adjacent cloud point;

[0018] Obtain the distance between each first cloud point and its corresponding neighboring cloud points, and mark it as the neighboring distance.

[0019] Furthermore, obtaining dynamic point cloud regions based on continuous two-dimensional point cloud maps also includes the following sub-steps:

[0020] Obtain the adjacent distances of the second number of historically occurring flame segments and mark them as historical flame distances;

[0021] Obtain the range of historical flame distances and mark it as the flame movement range;

[0022] Determine whether each adjacent distance is within the range of flame movement. If so, mark the corresponding first cloud point as the point to be analyzed.

[0023] Furthermore, obtaining dynamic point cloud regions based on continuous two-dimensional point cloud maps also includes the following sub-steps:

[0024] Obtain any point to be analyzed and mark it as the starting analysis point. Establish an analysis region of a×a centered on the starting analysis point and mark it as the initial division region.

[0025] Determine if the initial partitioned region contains a new point to be analyzed. If not, continue searching for the next point to be analyzed as the starting point. If so, establish an a×a analysis region centered on the new point to be analyzed and mark it as the search partitioned region. Determine again if the search partitioned region contains a new point to be analyzed. Repeat the process of establishing a new search partitioned region and making the determination until the new search partitioned region does not contain a new point to be analyzed. Mark the continuously searched points to be analyzed as consecutive group analysis points and use the consecutive group analysis points from each search as a set of analysis data.

[0026] Furthermore, obtaining dynamic point cloud regions based on continuous two-dimensional point cloud maps also includes the following sub-steps:

[0027] In a set of analytical data, a continuous set of analytical points with a minimum and a maximum x-coordinate is obtained and marked as the first analytical point and the second analytical point, respectively. Similarly, a continuous set of analytical points with a minimum and a maximum y-coordinate is obtained and marked as the third analytical point and the fourth analytical point, respectively. Then, line segments passing through the first analytical point and parallel to the Y-axis, the second analytical point and parallel to the Y-axis, the third analytical point and parallel to the X-axis, and the fourth analytical point and parallel to the X-axis are obtained and marked as the first line segment, the second line segment, the third line segment, and the fourth line segment, respectively. The rectangular area formed by the first, second, third, and fourth line segments is marked as the dynamic point cloud area.

[0028] Further, converting the historical flame image to grayscale to obtain a historical flame grayscale image includes the following sub-steps:

[0029] Obtain the RGB value of each pixel in the historical flame image and mark it as the historical flame RGB value;

[0030] The grayscale image of historical flames is obtained by converting all the RGB values ​​of historical flames in the historical flame image to grayscale values ​​using a grayscale conversion formula.

[0031] Furthermore, obtaining the flame grayscale threshold range based on historical flame grayscale images includes the following sub-steps:

[0032] Mark the grayscale values ​​of pixels in the historical flame grayscale image as flame grayscale values;

[0033] Divide all the ranges of flame grayscale values ​​into b equal ranges, and mark them as the flame grayscale division ranges;

[0034] Count the frequency of each flame grayscale range and mark it as the flame grayscale frequency.

[0035] A histogram is drawn with the flame grayscale value on the X-axis, the flame grayscale frequency on the Y-axis, and the flame grayscale range as the histogram interval. This histogram is then labeled as the flame grayscale histogram.

[0036] Furthermore, obtaining the flame grayscale threshold range based on historical flame grayscale images includes the following sub-steps:

[0037] Obtain the sum of the grayscale frequencies and mark it as the total historical flame frequency;

[0038] The first frequency threshold is calculated as: C1 = 0.1 × D1 / b; where C1 is the first frequency threshold and D1 is the total historical flame frequency.

[0039] In the flame grayscale histogram, starting from the leftmost division of the flame grayscale frequency, we check to the right whether the division of the flame grayscale frequency is less than the first frequency threshold. If it is, we delete the part of the flame grayscale histogram corresponding to the division of the flame grayscale frequency, until it is no longer, and then we obtain the minimum value of the horizontal axis of the flame grayscale histogram after deletion, and mark it as the minimum value of the filter.

[0040] In the flame grayscale histogram, starting from the rightmost division of flame grayscale frequency, we check to the left whether the division of flame grayscale frequency is less than the first frequency threshold. If it is, we delete the part of the flame grayscale histogram corresponding to the division of flame grayscale frequency, until it stops when it is not. We then obtain the maximum value of the horizontal axis of the flame grayscale histogram after deletion and mark it as the maximum value to be filtered.

[0041] The range from the minimum to the maximum value to be filtered is marked as the flame grayscale threshold range.

[0042] Furthermore, based on the flame grayscale threshold range and the detection map, it is determined whether the dynamic point cloud region is an abnormal region. If so, a flame generation signal is issued, including the following steps:

[0043] The detection image is converted to grayscale to obtain a detection grayscale image; the grayscale value of each pixel in the detection grayscale image is obtained and marked as the detection grayscale value;

[0044] Calculate the mean of all detected grayscale values ​​and mark it as the mean of detected grayscale.

[0045] If the average grayscale value is within the range of the flame grayscale threshold, the dynamic point cloud region is identified as an abnormal region, and a flame generation signal is issued.

[0046] This application also provides a lidar smoke and fire monitoring visualization analysis system for forest protection, including: a point cloud map acquisition module, a region acquisition module, an image acquisition module, a grayscale processing module, a threshold acquisition module, and an anomaly judgment module;

[0047] The point cloud map acquisition module is used to continuously acquire forest point cloud maps through lidar, acquire a two-dimensional projection map of the forest point cloud map, and mark it as a two-dimensional point cloud map;

[0048] The region acquisition module is used to acquire dynamic point cloud regions based on continuous two-dimensional point cloud maps.

[0049] The image acquisition module is used to acquire an image of a dynamic point cloud region in the projection direction of the two-dimensional point cloud map, and mark it as a detection image;

[0050] The grayscale processing module is used to acquire a first number of historical flame images and perform grayscale processing on the historical flame images to obtain historical flame grayscale images.

[0051] The threshold acquisition module is used to obtain the flame grayscale threshold range based on historical flame grayscale images;

[0052] The anomaly detection module is used to determine whether the dynamic point cloud region is an abnormal region based on the flame grayscale threshold range and the detection map. If so, it sends a signal to generate a flame.

[0053] The beneficial effects of this invention are as follows: This invention obtains a dynamic point cloud region based on a continuous two-dimensional point cloud map, acquires an image of the dynamic point cloud region in the projection direction of the two-dimensional point cloud map, marks it as a detection map, acquires a first number of historical flame images, performs grayscale processing on the historical flame images to obtain historical flame grayscale images, obtains a flame grayscale threshold range based on the historical flame grayscale images, and determines whether the dynamic point cloud region is an abnormal region based on the flame grayscale threshold range and the detection map. If so, a flame generation signal is issued. The advantage is that it combines lidar detection and image detection for flame judgment, making flame judgment more accurate.

[0054] The present invention obtains a dynamic point cloud region, which has the advantage of initially identifying the dynamic point cloud region as a flame region. Subsequent image processing of the dynamic point cloud region reduces the amount of subsequent image processing and increases the judgment efficiency. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the system of the present invention;

[0056] Figure 2 This is a schematic diagram of the adjacent distances in this invention;

[0057] Figure 3 This is a schematic diagram of the search area defined in this invention;

[0058] Figure 4 This is a schematic diagram of the dynamic point cloud region of the present invention;

[0059] Figure 5 This is a schematic diagram of the flame grayscale histogram of the present invention;

[0060] Figure 6 This is a schematic diagram illustrating the minimum and maximum values ​​to be filtered according to the present invention.

[0061] Figure 7 This is a flowchart of the steps of the method of the present invention. Detailed Implementation

[0062] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0063] Example 1, please refer to Figure 1 As shown, this application provides a lidar-based smoke and fire monitoring and visualization analysis system for forest protection, comprising:

[0064] The point cloud acquisition module is used to continuously acquire forest point cloud maps using lidar, obtain a two-dimensional projection map of the forest point cloud map, and mark it as a two-dimensional point cloud map;

[0065] The region acquisition module is used to acquire dynamic point cloud regions based on continuous two-dimensional point cloud maps;

[0066] The region acquisition module is configured with an adjacent distance acquisition strategy, which includes:

[0067] The real-time acquired 2D point cloud map is marked as the first point cloud map, and the 2D point cloud map acquired in the previous frame is marked as the second point cloud map;

[0068] Establish a Cartesian coordinate system, labeled as the cloud point coordinate system. Place the first cloud map in the first quadrant and place the second cloud map in the same position as the first cloud map.

[0069] Mark each cloud point in the first cloud map as the first cloud point;

[0070] Mark each cloud point in the second point cloud map as a second cloud point;

[0071] In the cloud point coordinate system, the second cloud point that is closest to the first cloud point is marked as the adjacent cloud point;

[0072] Obtain the distance between each first cloud point and its corresponding neighboring cloud points, and mark them as adjacent distances;

[0073] For practical applications, please refer to Figure 2 As shown, obtain all adjacent distances, for example, adjacent distances of 0.31cm, 0.22cm and 0.51cm.

[0074] The region acquisition module is configured with a strategy for acquiring points to be analyzed. The strategies for acquiring points to be analyzed include:

[0075] Obtain the adjacent distances of the second number of historically occurring flame segments and mark them as historical flame distances;

[0076] Obtain the range of historical flame distances and mark it as the flame movement range;

[0077] Determine whether each adjacent distance is within the range of flame movement; if so, mark its corresponding first cloud point as the point to be analyzed.

[0078] In practical applications, for example, if the historical flame distance is obtained as 0.62cm, 0.11cm, and 0.32cm, then the flame movement range is obtained as 0.21cm to 0.62cm. The principle is that if the point cloud map is obtained before and after it is stationary, the position of each cloud point remains unchanged, that is, the adjacent distance is 0. If a flame is generated, the point cloud map obtained at this time changes, so the adjacent distance becomes larger.

[0079] The region acquisition module is configured with a coherent group analysis point acquisition strategy, which includes:

[0080] Obtain any point to be analyzed and mark it as the starting analysis point. Establish an analysis region of a×a centered on the starting analysis point and mark it as the initial division region.

[0081] Determine if the initial partitioned region contains a new point to be analyzed. If not, continue searching for the next point to be analyzed as the starting point. If so, establish an a×a analysis region centered on the new point to be analyzed and mark it as the search partitioned region. Determine again if the search partitioned region contains a new point to be analyzed. Then repeat the process of establishing a new search partitioned region and making the determination until the new search partitioned region does not contain a new point to be analyzed. Mark the continuously searched points to be analyzed as consecutive group analysis points and use the consecutive group analysis points from each search as a set of analysis data.

[0082] For practical applications, please refer to Figure 3 As shown, the setting 'a' can be based on the distribution of cloud points. For example, if the average interval of cloud points in the point cloud map is 0.5cm, 'a' can be set to twice the average interval of cloud points in the point cloud map, which is 1cm. This search method can eliminate individual noise points and obtain each changing fire area. A set of analysis data represents a changing area.

[0083] The region acquisition module is configured with a dynamic point cloud region acquisition strategy, which includes:

[0084] In a set of analytical data, a continuous set of analytical points with a minimum and a maximum x-coordinate is obtained and marked as the first analytical point and the second analytical point, respectively. A continuous set of analytical points with a minimum and a maximum y-coordinate is obtained and marked as the third analytical point and the fourth analytical point, respectively. Then, line segments passing through the first analytical point and parallel to the Y-axis, the second analytical point and parallel to the Y-axis, the third analytical point and parallel to the X-axis, and the fourth analytical point and parallel to the X-axis are obtained and marked as the first line segment, the second line segment, the third line segment, and the fourth line segment, respectively. The rectangular area formed by the first line segment, the second line segment, the third line segment, and the fourth line segment is marked as the dynamic point cloud area.

[0085] For practical applications, please refer to Figure 4 As shown, a dynamic point cloud region is obtained, which may be the region where flames are generated.

[0086] The image acquisition module is used to acquire images of dynamic point cloud regions in the projection direction of a two-dimensional point cloud map and mark them as detection maps;

[0087] The grayscale processing module is used to obtain a first number of historical flame images and perform grayscale processing on the historical flame images to obtain historical flame grayscale images.

[0088] The grayscale processing module is configured with grayscale processing strategies, which include:

[0089] Obtain the RGB value of each pixel in the historical flame image and mark it as the historical flame RGB value;

[0090] The grayscale image of historical flames is obtained by converting all the RGB values ​​of historical flames in the historical flame image to grayscale values ​​using a grayscale conversion formula.

[0091] The threshold acquisition module is used to obtain the range of flame grayscale thresholds based on historical flame grayscale images;

[0092] The threshold acquisition module is configured with a flame grayscale histogram building strategy, which includes:

[0093] Mark the grayscale values ​​of pixels in the historical flame grayscale image as flame grayscale values;

[0094] Divide all the ranges of flame grayscale values ​​into b equal ranges, and mark them as the flame grayscale division ranges;

[0095] Count the frequency of each flame grayscale range and mark it as the flame grayscale frequency.

[0096] A histogram is drawn with the flame grayscale value on the X-axis, the flame grayscale frequency on the Y-axis, and the flame grayscale range as the histogram interval. This histogram is then marked as the flame grayscale histogram.

[0097] For practical applications, please refer to Figure 5 As shown, the range of all flame grayscale values ​​is divided into 7 equal ranges to obtain the flame grayscale histogram.

[0098] The threshold acquisition module is configured with a first frequency threshold acquisition strategy, which includes:

[0099] Obtain the sum of the grayscale frequencies and mark it as the total historical flame frequency;

[0100] The first frequency threshold is calculated as: C1 = 0.1 × D1 / b; where C1 is the first frequency threshold and D1 is the total historical flame frequency; it is set to 0.1 because 0.1 is a small proportion, which is considered to be a low distribution.

[0101] For practical applications, please refer to Figure 5 As shown, the total historical flame frequency is 4.05 million. The first frequency threshold is: C1 = 0.1 × 4.05 / 7 = 58,000, and the result is rounded to one decimal place.

[0102] In the flame grayscale histogram, starting from the leftmost division of the flame grayscale frequency, we check to the right whether the division of the flame grayscale frequency is less than the first frequency threshold. If it is, we delete the part of the flame grayscale histogram corresponding to the division of the flame grayscale frequency, until it is no longer, and then we obtain the minimum value of the horizontal axis of the flame grayscale histogram after deletion, and mark it as the minimum value of the filter.

[0103] In the flame grayscale histogram, starting from the rightmost division of flame grayscale frequency, we check to the left whether the division of flame grayscale frequency is less than the first frequency threshold. If it is, we delete the part of the flame grayscale histogram corresponding to the division of flame grayscale frequency, until it stops when it is not. We then obtain the maximum value of the horizontal axis of the flame grayscale histogram after deletion and mark it as the maximum value to be filtered.

[0104] The range from the minimum to the maximum value to be filtered is marked as the flame grayscale threshold range;

[0105] For practical applications, please refer to Figure 5 and Figure 6As shown, in the flame grayscale histogram, starting from the leftmost division of the flame grayscale frequency of 10,000, we judge to the right if the flame grayscale frequency of the division is less than 58,000. We delete the part of the flame grayscale histogram corresponding to the division of the flame grayscale frequency of 110,000 until it is greater than 58,000. The minimum value of the horizontal axis of the flame grayscale histogram at this time is 225, so the minimum value is 225. Starting from the rightmost division of the flame grayscale frequency of 750,000, we judge to the left if the flame grayscale frequency of the division is greater than 58,000, and we stop judging. The maximum value of the horizontal axis of the flame grayscale histogram at this time is 255, so the maximum value is 255. By filtering the less distributed data, we obtain a more accurate flame grayscale threshold range.

[0106] The anomaly detection module is used to determine whether a dynamic point cloud region is an abnormal region based on the flame grayscale threshold range and the detection map. If so, it sends a signal to generate a flame.

[0107] The exception detection module is configured with exception detection strategies, which include:

[0108] The detection image is converted to grayscale to obtain a detection grayscale image; the grayscale value of each pixel in the detection grayscale image is obtained and marked as the detection grayscale value;

[0109] Calculate the mean of all detected grayscale values ​​and mark it as the mean of detected grayscale.

[0110] If the average grayscale value is within the range of the flame grayscale threshold, the dynamic point cloud area is identified as an abnormal area, and a flame generation signal is issued.

[0111] In practical applications, if the average grayscale value of the detected area is 234, which is within the range of 225 to 255 of the flame grayscale threshold, then the dynamic point cloud area is considered to be an abnormal area. Therefore, a flame generation signal is issued, indicating that the dynamic point cloud area has changed and the grayscale value is close to that of the flame when it changes.

[0112] Example 2, please refer to Figure 7 As shown, this application provides a laser radar-based visual analysis method for monitoring and analyzing smoke and fire in forest protection, comprising the following steps:

[0113] Step S1: Continuously acquire forest point cloud maps using lidar, obtain a two-dimensional projection map of the forest point cloud maps, and mark it as a two-dimensional point cloud map.

[0114] Step S2: Obtain the dynamic point cloud region based on the continuous two-dimensional point cloud map; Step S2 includes the following sub-steps:

[0115] Step S201: Mark the real-time acquired two-dimensional point cloud map as the first point cloud map, and mark the two-dimensional point cloud map acquired in the previous frame as the second point cloud map;

[0116] Step S202: Establish a Cartesian coordinate system, labeled as the cloud point coordinate system, place the first point cloud map in the first quadrant, and place the second point cloud map in the same position as the first point cloud map;

[0117] Step S203: Mark each cloud point in the first point cloud map as a first cloud point; mark each cloud point in the second point cloud map as a second cloud point;

[0118] Step S204: In the cloud point coordinate system, mark the second cloud point that is closest to the first cloud point as an adjacent cloud point;

[0119] Step S205: Obtain the distance between each first cloud point and its corresponding adjacent cloud point, and mark it as the adjacent distance.

[0120] Step S206: Obtain the adjacent distances of the second number of historically occurring flame portions and mark them as historical flame distances;

[0121] Step S207: Obtain the range of historical flame distances and mark it as the flame movement range;

[0122] Step S208: Determine whether each adjacent distance is within the range of flame movement. If so, mark the corresponding first cloud point as the point to be analyzed.

[0123] Step S209: Obtain any point to be analyzed, mark it as the starting analysis point, and establish an analysis region of a×a centered on the starting analysis point, and mark it as the starting division region;

[0124] Step S210: Determine whether the initial partitioned region contains a new point to be analyzed. If not, continue to search for the next point to be analyzed as the starting point for analysis. If so, establish an analysis region of size a×a with the new point to be analyzed as the center and mark it as the search partitioned region. Determine again whether the search partitioned region contains a new point to be analyzed. Then repeat the process of establishing a new search partitioned region and making the determination until the new search partitioned region does not contain a new point to be analyzed. Mark the continuously searched points to be analyzed as consecutive group analysis points and use the consecutive group analysis points searched each time as a set of analysis data.

[0125] Step S211: Under a set of analysis data, obtain a continuous set of analysis points with a minimum and a maximum x-coordinate, and mark them as the first analysis point and the second analysis point, respectively. Obtain a continuous set of analysis points with a minimum and a maximum y-coordinate, and mark them as the third analysis point and the fourth analysis point, respectively. Then, obtain the line segments passing through the first analysis point and parallel to the Y-axis, the line segments passing through the second analysis point and parallel to the Y-axis, the line segments passing through the third analysis point and parallel to the X-axis, and the line segments passing through the fourth analysis point and parallel to the X-axis, and mark them as the first line segment, the second line segment, the third line segment, and the fourth line segment, respectively. The rectangular area formed by the first line segment, the second line segment, the third line segment, and the fourth line segment is marked as the dynamic point cloud area.

[0126] Step S3: Obtain an image of the dynamic point cloud region in the projection direction of the two-dimensional point cloud map and mark it as a detection map.

[0127] Step S4: Obtain a first number of historical flame images, and perform grayscale processing on the historical flame images to obtain historical flame grayscale images; Step S4 includes the following sub-steps:

[0128] Step S401: Obtain the RGB value of each pixel in the historical flame image and mark it as the historical flame RGB value;

[0129] Step S402: Use the grayscale conversion formula to convert all the historical flame RGB values ​​in the historical flame image into grayscale values ​​to obtain the historical flame grayscale image.

[0130] Step S5: Obtain the flame grayscale threshold range based on historical flame grayscale images; Step S5 includes the following sub-steps:

[0131] Step S501: Mark the grayscale values ​​of the pixels in the historical flame grayscale image as flame grayscale values;

[0132] Step S502: Divide all the ranges of flame grayscale values ​​into b equal ranges, and mark them as the flame grayscale division ranges;

[0133] Step S503: Count the frequency of each flame grayscale division range and mark it as the frequency of flame grayscale division;

[0134] Step S504: Using the flame grayscale value as the X-axis, the flame grayscale frequency as the Y-axis, and the flame grayscale range as the histogram interval, draw a histogram and mark it as the flame grayscale histogram.

[0135] Step S505: Obtain the sum of the grayscale frequencies and mark it as the total historical flame frequency;

[0136] Step S506: Calculate the first frequency threshold as: C1 = 0.1 × D1 / b; where C1 is the first frequency threshold and D1 is the total historical flame frequency.

[0137] Step S507: In the flame grayscale histogram, starting from the leftmost division of flame grayscale frequency in the flame grayscale histogram, determine whether the division of flame grayscale frequency is less than the first frequency threshold. If it is, delete the part of the flame grayscale histogram corresponding to the division of flame grayscale frequency until it is not, and then obtain the minimum value of the horizontal axis of the flame grayscale histogram after deletion, and mark it as the minimum value for filtering.

[0138] Step S508: In the flame grayscale histogram, starting from the rightmost division of flame grayscale frequency in the flame grayscale histogram, determine whether the division of flame grayscale frequency is less than the first frequency threshold. If so, delete the part of the flame grayscale histogram corresponding to the division of flame grayscale frequency until it is not, and then obtain the maximum value of the horizontal axis of the flame grayscale histogram after deletion, and mark it as the maximum value to be filtered.

[0139] Step S509: Mark the range from the minimum to the maximum value of the filter as the flame grayscale threshold range.

[0140] Step S6: Based on the flame grayscale threshold range and the detection map, determine whether the dynamic point cloud region is an abnormal region. If so, issue a flame generation signal. Step S6 includes the following sub-steps:

[0141] Step S601: Convert the detection image to grayscale to obtain a detection grayscale image; obtain the grayscale value of each pixel in the detection grayscale image and mark it as the detection grayscale value;

[0142] Step S602: Calculate the mean of all detected grayscale values ​​and mark it as the mean of detected grayscale.

[0143] Step S603: Obtain whether the average grayscale value of the detection is within the grayscale threshold range of the flame. If so, the dynamic point cloud area is identified as an abnormal area, and a flame generation signal is issued.

[0144] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0145] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings or direct couplings or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

Claims

1. A laser radar-based visual analysis method for monitoring and analyzing smoke and fire in forest protection, characterized in that, Includes the following steps: Forest point cloud images are continuously acquired using lidar, and a two-dimensional projection of the forest point cloud images is obtained and marked as a two-dimensional point cloud image. Dynamic point cloud regions are obtained based on continuous two-dimensional point cloud maps; The image of the dynamic point cloud region is obtained in the projection direction of the two-dimensional point cloud map and marked as the detection map; Obtain the first number of historical flame images, and perform grayscale processing on the historical flame images to obtain historical flame grayscale images; Obtain the flame grayscale threshold range based on historical flame grayscale images; Based on the flame grayscale threshold range and the detection map, determine whether the dynamic point cloud region is an abnormal region. If so, issue a signal to generate a flame. Obtaining a dynamic point cloud region based on a continuous two-dimensional point cloud map includes the following sub-steps: The real-time acquired 2D point cloud map is marked as the first point cloud map, and the 2D point cloud map acquired in the previous frame is marked as the second point cloud map; Establish a Cartesian coordinate system, labeled as the cloud point coordinate system. Place the first cloud map in the first quadrant and place the second cloud map in the same position as the first cloud map. Mark each cloud point in the first cloud map as the first cloud point; Mark each cloud point in the second point cloud map as a second cloud point; In the cloud point coordinate system, the second cloud point that is closest to the first cloud point is marked as the adjacent cloud point; Obtain the distance between each first cloud point and its corresponding neighboring cloud points, and mark them as adjacent distances; Obtaining a dynamic point cloud region based on a continuous two-dimensional point cloud map also includes the following sub-steps: Obtain the adjacent distances of the second number of historically occurring flame segments and mark them as historical flame distances; Obtain the range of historical flame distances and mark it as the flame movement range; Determine whether each adjacent distance is within the range of flame movement; if so, mark its corresponding first cloud point as the point to be analyzed. Obtaining a dynamic point cloud region based on a continuous two-dimensional point cloud map also includes the following sub-steps: Obtain any point to be analyzed and mark it as the starting analysis point. Establish an analysis region of a×a centered on the starting analysis point and mark it as the initial division region. Determine if the initial partitioned region contains a new point to be analyzed. If not, continue searching for the next point to be analyzed as the starting point. If so, establish an a×a analysis region centered on the new point to be analyzed and mark it as the search partitioned region. Determine again if the search partitioned region contains a new point to be analyzed. Then repeat the process of establishing a new search partitioned region and making the determination until the new search partitioned region does not contain a new point to be analyzed. Mark the continuously searched points to be analyzed as consecutive group analysis points and use the consecutive group analysis points from each search as a set of analysis data. Obtaining a dynamic point cloud region based on a continuous two-dimensional point cloud map also includes the following sub-steps: In a set of analytical data, a continuous set of analytical points with a minimum and a maximum x-coordinate is obtained and marked as the first analytical point and the second analytical point, respectively. Similarly, a continuous set of analytical points with a minimum and a maximum y-coordinate is obtained and marked as the third analytical point and the fourth analytical point, respectively. Then, line segments passing through the first analytical point and parallel to the Y-axis, the second analytical point and parallel to the Y-axis, the third analytical point and parallel to the X-axis, and the fourth analytical point and parallel to the X-axis are obtained and marked as the first line segment, the second line segment, the third line segment, and the fourth line segment, respectively. The rectangular area formed by the first, second, third, and fourth line segments is marked as the dynamic point cloud area.

2. The laser radar smoke and fire monitoring visualization analysis method for forest protection according to claim 1, characterized in that, The process of converting historical flame images to grayscale to obtain historical flame grayscale images includes the following sub-steps: Obtain the RGB value of each pixel in the historical flame image and mark it as the historical flame RGB value; The grayscale image of historical flames is obtained by converting all the RGB values ​​of historical flames in the historical flame image to grayscale values ​​using a grayscale conversion formula.

3. The laser radar smoke and fire monitoring visualization analysis method for forest protection according to claim 2, characterized in that, Obtaining the flame grayscale threshold range based on historical flame grayscale images includes the following sub-steps: Mark the grayscale values ​​of pixels in the historical flame grayscale image as flame grayscale values; Divide all the ranges of flame grayscale values ​​into b equal ranges, and mark them as the flame grayscale division ranges; Count the frequency of each flame grayscale range and mark it as the flame grayscale frequency. A histogram is drawn with the flame grayscale value on the X-axis, the flame grayscale frequency on the Y-axis, and the flame grayscale range as the histogram interval. This histogram is then labeled as the flame grayscale histogram.

4. The laser radar smoke and fire monitoring visualization analysis method for forest protection according to claim 3, characterized in that, Obtaining the flame grayscale threshold range based on historical flame grayscale images includes the following sub-steps: Obtain the sum of the grayscale frequencies and mark it as the total historical flame frequency; The first frequency threshold is calculated as: C1 = 0.1 × D1 / b; where C1 is the first frequency threshold and D1 is the total historical flame frequency. In the flame grayscale histogram, starting from the leftmost division of the flame grayscale frequency, we check to the right whether the division of the flame grayscale frequency is less than the first frequency threshold. If it is, we delete the part of the flame grayscale histogram corresponding to the division of the flame grayscale frequency, until it is no longer, and then we obtain the minimum value of the horizontal axis of the flame grayscale histogram after deletion, and mark it as the minimum value of the filter. In the flame grayscale histogram, starting from the rightmost division of flame grayscale frequency, we check to the left whether the division of flame grayscale frequency is less than the first frequency threshold. If it is, we delete the part of the flame grayscale histogram corresponding to the division of flame grayscale frequency, until it stops when it is not. We then obtain the maximum value of the horizontal axis of the flame grayscale histogram after deletion and mark it as the maximum value to be filtered. The range from the minimum to the maximum value to be filtered is marked as the flame grayscale threshold range.

5. The laser radar smoke and fire monitoring visualization analysis method for forest protection according to claim 4, characterized in that, Based on the flame grayscale threshold range and the detection map, it is determined whether the dynamic point cloud region is an abnormal region. If so, a flame generation signal is issued, including the following steps: The detection image is converted to grayscale to obtain a detection grayscale image; the grayscale value of each pixel in the detection grayscale image is obtained and marked as the detection grayscale value; Calculate the mean of all detected grayscale values ​​and mark it as the mean of detected grayscale. If the average grayscale value is within the range of the flame grayscale threshold, the dynamic point cloud region is identified as an abnormal region, and a flame generation signal is issued.

6. A lidar-based visual analysis system for monitoring and analyzing smoke and fire in forest protection, used to implement the lidar-based visual analysis method for monitoring and analyzing smoke and fire in forest protection as described in any one of claims 1-5, characterized in that, It includes a point cloud acquisition module, a region acquisition module, an image acquisition module, a grayscale processing module, a threshold acquisition module, and an anomaly detection module; The point cloud map acquisition module is used to continuously acquire forest point cloud maps through lidar, acquire a two-dimensional projection map of the forest point cloud map, and mark it as a two-dimensional point cloud map; The region acquisition module is used to acquire dynamic point cloud regions based on continuous two-dimensional point cloud maps. The image acquisition module is used to acquire an image of a dynamic point cloud region in the projection direction of the two-dimensional point cloud map, and mark it as a detection image; The grayscale processing module is used to acquire a first number of historical flame images and perform grayscale processing on the historical flame images to obtain historical flame grayscale images. The threshold acquisition module is used to obtain the flame grayscale threshold range based on historical flame grayscale images; The anomaly detection module is used to determine whether the dynamic point cloud region is an abnormal region based on the flame grayscale threshold range and the detection map. If so, it sends a signal to generate a flame.

Images