Image data-based mutual feedback type geological disaster detection system and method

Abstract

The invention relates to the technical field of geological disaster detection, in particular to a mutual feedback type geological disaster detection system and method based on image data. The system comprises a detection marking unit, a detection judgment unit and a sorting selection unit, the detection judgment unit feeds back three-dimensional coordinates to an unmanned aerial vehicle, and the unmanned aerial vehicle detects rock textures according to the three-dimensional coordinates fed back by the three-dimensional model of the hilly land shape to obtain point cloud data of the rock textures. And judging whether unclear rock textures exist or not according to the detected rock texture point cloud data, and removing the unclear rock textures. The data judgment module performs re-judgment by using the distance difference data and the intensity difference data, re-judges whether the removed rock texture point cloud data has the rainwater flower abnormal feature points or not, re-removes the rainwater flower abnormal feature points according to the re-judged abnormal feature points, and determines whether the rainwater flower abnormal feature points exist in the rock texture point cloud data or not by re-removing the rainwater flower abnormal feature points. The method is used for more accurately identifying geological features of cracks and landslide traces of rocks, so that the detection precision of geological disasters is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of geological disaster detection, and in particular to a geological disaster detection system and method based on image data mutual feedback. Background Art

[0002] Geological disaster detection is a technology that integrates multiple technical means to achieve real-time monitoring, early warning and assessment of geological disasters through the collection, processing and analysis of image data. It also combines multiple technologies such as drone aerial photography, remote monitoring and image processing to improve the efficiency and accuracy of geological disaster detection and reduce the losses and impacts caused by disasters.

[0003] When it rains around the hillside terrain, drones are used to detect the rock texture in the rain, and the detected rocks are located at the potential points of debris flow geological disasters. When it rains, raindrops falling on the rock surface will produce raindrops, which reduce the quality of the detected rock texture data, thereby covering up or misleading the texture features of the rock surface, resulting in a decrease in the accuracy of the detected debris flow geological disaster potential points. In order to avoid reducing the accuracy of detecting debris flow geological disaster potential points due to rain, we provide a geological disaster detection system and method based on image data mutual feedback. Summary of the invention

[0004] The purpose of the present invention is to provide a geological disaster detection system and method based on image data mutual feedback to solve the problems raised in the above background technology.

[0005] To achieve the above-mentioned object, one of the objects of the present invention is to provide a geological disaster detection system based on image data mutual feedback, including a detection marking unit, a detection judgment unit, and a sorting and selection unit;

[0006] The detection and marking unit detects the hillside terrain using a drone and generates a three-dimensional model of the hillside terrain, collects rock texture image data based on the detected point cloud data, and inputs the collected rock texture image data into the three-dimensional model;

[0007] The detection and judgment unit is used to receive data from the detection and marking unit. The three-dimensional model of the hillside terrain feeds back the three-dimensional coordinates to the drone. The drone detects the rock texture according to the three-dimensional coordinates fed back by the three-dimensional model of the hillside terrain, obtains the point cloud data of the rock texture, and judges whether there is unclear rock texture according to the detected rock texture point cloud data and removes it.

[0008] The detection and marking unit receives the rock texture point cloud data after being eliminated by the detection and judgment unit, calculates the distance difference and the intensity difference, and uses the calculated distance difference and intensity difference data to make a second judgment, and again judges whether there are abnormal feature points of raindrops in the rock texture point cloud data after being eliminated and eliminates them again;

[0009] The sorting and selection unit is used to receive the rock texture point cloud data after re-elimination from the detection and judgment unit and the historical data of the detection and marking unit, and sort them in sequence from low to high according to the laser echo intensity corresponding to the rock texture point cloud data after re-elimination, select the rock texture point cloud data after elimination with the highest laser echo intensity, and transmit the selected rock texture point cloud data to the detection and marking unit. The detection and marking unit compares the selected rock texture point cloud data of the same three-dimensional coordinate point with the collected rock texture image data to compare whether there is a difference in texture, and identifies the three-dimensional coordinate point as a geological disaster risk point according to the comparison result.

[0010] As a further improvement of the technical solution, the detection and marking unit includes a detection modeling module and a hidden danger point marking module;

[0011] The detection and modeling module is used to obtain the rainfall time and historical data of the hillside in real time. 15 minutes before the rainfall on the hillside, a drone equipped with a laser scanner is used to detect the hillside terrain and obtain point cloud data. The point cloud data contains the details of the rock texture and three-dimensional coordinate information. The rock texture image is collected according to the detected point cloud data, and the corresponding three-dimensional coordinates are recorded. At the same time, the detected point cloud data is densely interpolated and a three-dimensional model of the hillside terrain is generated;

[0012] The historical data include point cloud density and laser echo intensity corresponding to rock texture point cloud data;

[0013] The hidden danger point marking module is used to receive the rock texture image data, corresponding three-dimensional coordinates, and three-dimensional model of the hillside terrain collected in the detection and modeling module, mark the three-dimensional model of the hillside terrain according to the corresponding three-dimensional coordinates, and input the collected rock texture image data into the three-dimensional model at the marked location.

[0014] As a further improvement of the technical solution, the detection and judgment unit includes a rock detection module and a data judgment module;

[0015] The rock detection module is used to receive the rainfall time of the hillside, the collected rock texture image data, the corresponding three-dimensional coordinates, and the three-dimensional model of the hillside terrain in the detection and modeling module. 15 minutes after the hillside rainfall and when the surrounding area begins to rain, the three-dimensional model of the hillside terrain will feed back the marked three-dimensional coordinates to the drone. The drone accurately locates the rock position according to the marked three-dimensional coordinates. The drone then uses the equipped laser scanner to detect the rock texture and obtain the point cloud data of the rock texture.

[0016] The data judgment module is used to receive the historical data of the detection and modeling module, receive the rock texture point cloud data detected by the rock detection module and the number of times the rock is detected, use the time window filtering method to filter the rock texture point cloud data, and then judge based on the laser echo intensity in the historical data and the point cloud density in the historical data to determine whether the detected rock texture point cloud data has unclear rock texture, eliminate the detected rock texture point cloud data based on the judged unclear rock texture data, and feed the eliminated rock texture point cloud data back to the three-dimensional model of the hillside terrain in the detection and modeling module.

[0017] As a further improvement of the technical solution, the implementation principle of judging whether the rock texture is unclear in the data judgment module is as follows:

[0018] First, collect the laser echo intensity I j and point cloud density ρ j And make a judgment to determine the detected rock texture point cloud data p j Whether there is unclear rock texture, and obtain the unclear rock texture data Pd ng , the specific algorithm formula is:

[0019]

[0020] Among them, I yz refers to the set standard intensity threshold, ρ yz′ Refers to the set standard density threshold, p j Refers to the jth rock texture point cloud data detected, I j refers to the laser echo intensity corresponding to the j-th rock texture point cloud data, ρ j Refers to the point cloud density corresponding to the j-th rock texture point cloud data;

[0021] When the laser echo intensity I j Less than the set standard intensity threshold I yz , or the point cloud density ρ j Less than the set standard density threshold ρ yz′ When Pd ng =1, indicating that the detected rock texture point cloud data p j There are unclear rock textures;

[0022] When the laser echo intensity I j Greater than or equal to the set standard intensity threshold I yz When, or the point cloud density ρ j Greater than or equal to the set standard density threshold ρ yz′ When Pd ng =0, indicating that the detected rock texture point cloud data p j There is no unclear rock texture.

[0023] As a further improvement of the present technical solution, the detection modeling module receives the rock texture point cloud data eliminated by the data judgment module, analyzes the point cloud data adjacent to the eliminated rock texture point cloud data based on the three-dimensional model data of the hillside terrain, calculates the distance difference between the eliminated rock texture point cloud data and the adjacent point cloud data, then takes the laser echo intensity corresponding to the eliminated rock texture point cloud data and the adjacent point cloud data from the historical data, calculates the intensity difference based on the laser echo intensity, and feeds the distance difference and intensity difference data back to the data judgment module.

[0024] As a further improvement of the present technical solution, the data judgment module receives the distance difference and intensity difference data in the detection modeling module, and uses the distance difference data and the intensity difference data to make a second judgment, and judges again whether there are abnormal feature points of raindrops in the rock texture point cloud data after elimination, and eliminates them again based on the abnormal feature points judged again, and obtains the rock texture point cloud data after elimination again.

[0025] As a further improvement of the technical solution, the implementation principle of whether the rock texture point cloud data after being eliminated again in the data judgment module has abnormal feature points of raindrops is as follows:

[0026] First, collect the distance difference data d (p,q) and intensity difference data Δn (p,q) And make a second judgment to obtain the rock texture point cloud data Zpd (p,q) , the specific algorithm formula is:

[0027]

[0028] Among them, d refers to the set standard distance threshold, and Δn refers to the set standard intensity difference threshold;

[0029] When the distance difference data d (p,q) Less than the set standard distance threshold d, and the intensity difference data Δn (p,q) When the intensity difference is less than the set standard intensity difference threshold Δn, Zpd (p,w) =1, indicating that there are abnormal feature points of rain water flowers in the rock texture point cloud data after elimination;

[0030] When the distance difference data d (p,w) When the distance is greater than or equal to the set standard distance threshold d, and the intensity difference data Δn (p,q) When it is greater than or equal to the set standard intensity difference threshold Δn, Pd ng =0, indicating that there are no abnormal feature points of raindrops in the rock texture point cloud data after elimination again.

[0031] As a further improvement of the present technical solution, the sorting and selection unit is used to receive the rock texture point cloud data after re-elimination of the data judgment module and the historical data of the detection and modeling module, obtain the laser echo intensity corresponding to the rock texture point cloud data after re-elimination from the historical data, and sort them in sequence from low to high according to the laser echo intensity, select the rock texture point cloud data after re-elimination with the highest laser echo intensity, and transmit the selected rock texture point cloud data to the hidden danger point marking module.

[0032] As a further improvement of the present technical solution, the module for marking hidden danger points receives the rock texture point cloud data selected in the sorting and selecting unit, inputs the selected rock texture point cloud data and the corresponding three-dimensional coordinates into the three-dimensional model data of the hillside terrain, and the three-dimensional model of the hillside terrain compares the selected rock texture point cloud data of the same three-dimensional coordinate point with the collected rock texture image data to compare whether there is a difference in texture, and identifies the three-dimensional coordinate point as a geological disaster hidden danger point according to the comparison result;

[0033] Specific comparison:

[0034] Case 1: When the selected rock texture point cloud data is different from the rock texture in the collected rock texture image data, it means that the rainfall has caused the rocks on the hillside to move and change, and the three-dimensional coordinate point corresponding to the selected rock texture point cloud data is a geological disaster risk point;

[0035] Case 2: When the selected rock texture point cloud data is the same as the rock texture in the collected rock texture image data, it means that the rainfall has not caused the rocks on the hillside to move or change, and the three-dimensional coordinate point corresponding to the selected rock texture point cloud data is not a geological disaster risk point.

[0036] The second object of the present invention is to provide a method for operating an image data mutual feedback type geological disaster detection system including any one of the above, comprising the following method steps:

[0037] S1, the detection and modeling module uses a UAV equipped with a laser scanner to detect the hillside terrain and obtain point cloud data, collects images of rock textures based on the detected point cloud data, and records the corresponding three-dimensional coordinates, and generates a three-dimensional model of the hillside terrain based on the detected point cloud data;

[0038] S2, the module for marking hidden danger points marks the three-dimensional model of the hillside terrain according to the corresponding three-dimensional coordinates, and inputs the collected rock texture image data into the three-dimensional model at the marked location;

[0039] S3, the rock detection module uses the drone to accurately locate the rock position according to the three-dimensional coordinates fed back by the three-dimensional model of the hillside terrain. The drone then uses the laser scanner equipped to detect the rock texture and obtain the point cloud data of the rock texture;

[0040] S4, the data judgment module filters the rock texture point cloud data according to the detected rock texture point cloud data and the number of rock detections, and then judges whether the detected rock texture point cloud data has unclear rock texture and removes it according to the laser echo intensity in the historical data and the point cloud density in the historical data;

[0041] S5, the detection modeling module receives the rock texture point cloud data after being eliminated by the data judgment module, and calculates the distance difference and the intensity difference, and then uses the distance difference data and the intensity difference data to make another judgment, and again judges whether there are abnormal feature points of rain water flowers in the rock texture point cloud data after being eliminated and eliminates them again;

[0042] S6. The sorting and selection unit is used to receive the rock texture point cloud data after the re-elimination from the data judgment module and sort them in sequence from low to high, and input the selected rock texture point cloud data and the corresponding three-dimensional coordinates into the three-dimensional model data of the hillside terrain. The three-dimensional model of the hillside terrain uses the selected rock texture point cloud data of the same three-dimensional coordinate point to compare with the collected rock texture image data to compare whether there is a difference in texture, and identifies the three-dimensional coordinate point as a geological disaster risk point according to the comparison result.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. In the geological disaster detection system and method based on image data mutual feedback, the data judgment module filters the rock texture point cloud data according to the detected rock texture point cloud data and the number of times the rock is detected, and then judges whether the detected rock texture point cloud data has unclear rock texture according to the laser echo intensity in the historical data and the point cloud density in the historical data. The detected rock texture point cloud data is eliminated according to the judged unclear rock texture data. By eliminating the unclear rock texture point cloud data, the noise and interference in the rock texture point cloud data are reduced, thereby improving the accuracy of geological disaster detection.

[0045] 2. In the geological disaster detection system and method based on image data mutual feedback, the data judgment module uses the distance difference data and the intensity difference data to make a second judgment, and judges again whether there are abnormal feature points of raindrops in the rock texture point cloud data after elimination, and eliminates them again according to the abnormal feature points judged again. By eliminating these abnormal feature points of raindrops again, it is used to more accurately identify the geological characteristics of rock cracks and landslide traces, thereby improving the detection accuracy of geological disasters.

[0046] 3. In the geological disaster detection system and method based on image data mutual feedback, the hidden danger point marking module inputs the selected rock texture point cloud data and the corresponding three-dimensional coordinates into the three-dimensional model data of the hillside terrain. The three-dimensional model of the hillside terrain uses the selected rock texture point cloud data of the same three-dimensional coordinate point to compare with the collected rock texture image data to compare whether there are texture differences. The three-dimensional coordinate point is identified as a geological disaster hidden danger point based on the comparison result. By identifying the difference in rock texture, it is possible to detect whether the rock has moved or changed early, thereby detecting potential geological disaster risk points, facilitating early warning and taking timely measures to reduce casualties and property losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is an overall block diagram of the present invention;

[0048] Figure 2 A block diagram of a detection tag unit of the present invention;

[0049] Figure 3 This is a block diagram of the detection and judgment unit of the present invention.

[0050] The meaning of each number in the figure is:

[0051] 1. Detection and marking unit; 11. Detection modeling module; 12. Hidden danger point marking module;

[0052] 2. Detection and judgment unit; 21. Rock detection module; 22. Data judgment module;

[0053] 3. Sort and select units. DETAILED DESCRIPTION

[0054] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0055] Example 1

[0056] See also Figure 1-Figure 3 As shown, one of the purposes of this embodiment is to provide a geological disaster detection system based on image data mutual feedback, including a detection marking unit 1, a detection judgment unit 2, and a sorting selection unit 3;

[0057] The detection and marking unit 1 uses a drone to detect the hillside terrain and generate a three-dimensional model of the hillside terrain, collects rock texture image data based on the detected point cloud data, and inputs the collected rock texture image data into the three-dimensional model; the detection and judgment unit 2 is used to receive the data in the detection and marking unit 1, and the three-dimensional model of the hillside terrain feeds back the three-dimensional coordinates to the drone. The drone detects the rock texture according to the three-dimensional coordinates fed back by the three-dimensional model of the hillside terrain, obtains the point cloud data of the rock texture, and judges whether there is unclear rock texture and removes it according to the detected rock texture point cloud data; the detection and marking unit 1 receives the rock texture point cloud data removed by the detection and judgment unit 2, and calculates the distance difference and intensity difference, and uses the calculated distance difference and intensity difference data to re- The rock texture point cloud data after elimination is judged again, and whether there are abnormal characteristic points of raindrops are eliminated again; the sorting and selection unit 3 is used to receive the rock texture point cloud data after elimination of the detection and judgment unit 2 and the historical data of the detection and marking unit 1, and sort them in sequence from low to high according to the laser echo intensity corresponding to the rock texture point cloud data after elimination, and select the rock texture point cloud data after elimination with the highest laser echo intensity, and transmit the selected rock texture point cloud data to the detection and marking unit 1. The detection and marking unit 1 compares the selected rock texture point cloud data of the same three-dimensional coordinate point with the collected rock texture image data to compare whether there are differences in texture, and identifies the three-dimensional coordinate point as a geological disaster risk point according to the comparison result.

[0058] The following is a refinement of the above units, see Figure 1-Figure 3 As shown;

[0059] The detection and marking unit 1 includes a detection modeling module 11 and a potential danger point marking module 12;

[0060] The detection and modeling module 11 is used to obtain the rainfall time and historical data of the hillside in real time. 15 minutes before the rainfall on the hillside, a drone equipped with a laser scanner is used to detect the hillside terrain and obtain point cloud data. The point cloud data contains the details of the rock texture and three-dimensional coordinate information. The rock texture image is collected according to the detected point cloud data, and the corresponding three-dimensional coordinates are recorded. At the same time, the detected point cloud data is densely interpolated and a three-dimensional model of the hillside terrain is generated. A high-resolution terrain model can be generated to capture more subtle terrain changes;

[0061] The historical data include the point cloud density corresponding to the rock texture point cloud data (the point cloud density is not fixed, but varies according to the location of the specific point in the point cloud data and the distribution of the surrounding points) and the laser echo intensity;

[0062] The hidden danger point marking module 12 is used to receive the rock texture image data, the corresponding three-dimensional coordinates, and the three-dimensional model of the hillside terrain collected in the detection modeling module 11, mark the three-dimensional model of the hillside terrain according to the corresponding three-dimensional coordinates, and input the collected rock texture image data into the three-dimensional model at the marked location;

[0063] The detection and judgment unit 2 includes a rock detection module 21 and a data judgment module 22;

[0064] The rock detection module 21 is used to receive the rainfall time of the hillside, the collected rock texture image data, the corresponding three-dimensional coordinates, and the three-dimensional model of the hillside terrain in the detection modeling module 11. 15 minutes after the hillside rains and when the surrounding area begins to rain, the three-dimensional model of the hillside terrain will feed back the marked three-dimensional coordinates to the drone. The drone accurately locates the rock position according to the marked three-dimensional coordinates. The drone then uses the equipped laser scanner to detect the rock texture and obtain the point cloud data of the rock texture.

[0065] The data judgment module 22 is used to receive the historical data of the detection and modeling module 11, receive the rock texture point cloud data detected by the rock detection module 21 and the number of times the rock is detected, use the time window filtering method to filter the rock texture point cloud data, and then judge based on the laser echo intensity in the historical data and the point cloud density in the historical data to determine whether the detected rock texture point cloud data has unclear rock texture, and eliminate the detected rock texture point cloud data based on the determined unclear rock texture data. Data with unclear texture may make it difficult to identify texture features. Eliminating these data can improve the accuracy of texture analysis, so that the finally generated model has higher texture recognition, and the eliminated rock texture point cloud data is fed back to the three-dimensional model of the hillside terrain in the detection and modeling module 11.

[0066] The implementation principle of filtering rock texture information using time window filtering method:

[0067] Collect the detected rock texture point cloud data Ys xxi The rock texture point cloud data is filtered based on the number of rock detection times N to obtain the filtered rock texture point cloud data Gl ysxx , the specific algorithm formula is:

[0068]

[0069] Among them, i represents the detected rock texture point cloud data Ys xxi The index in ;

[0070] The principle of judging that the rock texture is not clear:

[0071] First, collect the laser echo intensity I j and point cloud density ρj And make a judgment to determine the detected rock texture point cloud data p j Whether there is unclear rock texture, and obtain the unclear rock texture data Pd ng , the specific algorithm formula is:

[0072]

[0073] Among them, I yz refers to the set standard intensity threshold, ρ yz′ Refers to the set standard density threshold, p j Refers to the jth rock texture point cloud data detected, I j refers to the laser echo intensity corresponding to the j-th rock texture point cloud data, ρ j Refers to the point cloud density corresponding to the j-th rock texture point cloud data;

[0074] When the laser echo intensity I j Less than the set standard intensity threshold I yz , or the point cloud density ρ j Less than the set standard density threshold ρ yz′ When Pd ng =1, indicating that the detected rock texture point cloud data p j There are unclear rock textures;

[0075] When the laser echo intensity I j Greater than or equal to the set standard intensity threshold I yz When, or the point cloud density ρ j Greater than or equal to the set standard density threshold ρ yz′ When Pd ng =0, indicating that the detected rock texture point cloud data p j There is no unclear rock texture.

[0076] The detection modeling module 11 receives the rock texture point cloud data removed by the data judgment module 22, analyzes the point cloud data adjacent to the removed rock texture point cloud data according to the three-dimensional model data of the hillside terrain, calculates the distance difference between the removed rock texture point cloud data and the adjacent point cloud data, and then takes the laser echo intensity corresponding to the removed rock texture point cloud data and the adjacent point cloud data from the historical data, and calculates the intensity difference according to the laser echo intensity. Calculating the intensity difference can improve the classification accuracy of the point cloud data. For example, on a hillside, different rocks may have different reflection intensities. By analyzing these differences, the abnormal characteristic points of raindrops can be judged more accurately, and the distance difference and intensity difference data are fed back to the data judgment module 22;

[0077] Principle of calculating distance difference:

[0078] First, obtain the three-dimensional coordinates TP of the point cloud data after elimination n =(x p ,y p z p ) and the adjacent point cloud data 3D coordinates Xq m =(x q ,y q z q ) and calculate the distance difference to obtain the distance difference d (p,q) , the specific algorithm formula is:

[0079]

[0080] Among them, TP n Refers to the nth point cloud data after elimination, Xq m Refers to the mth adjacent point cloud data. This formula is used to calculate the distance difference. The precise matching of adjacent point cloud data can reduce the error, thereby improving the accuracy of distance calculation.

[0081] The implementation principle of calculating intensity difference:

[0082] Collect the laser echo intensity I corresponding to the point cloud data after elimination n The laser echo intensity I corresponding to the adjacent point cloud data m And calculate the intensity difference to get the intensity difference Δn (p,q) , the specific algorithm formula is:

[0083] Δn (p,q) =|I n -I m |;

[0084] Among them, I n Refers to the laser echo intensity corresponding to the nth point cloud data after elimination, I m Refers to the laser echo intensity corresponding to the adjacent m-th point cloud data;

[0085] The data judgment module 22 receives the distance difference and intensity difference data in the detection modeling module 11, and uses the distance difference data and the intensity difference data to make a second judgment, and judges whether there are raindrop abnormal feature points in the rock texture point cloud data after being eliminated, and eliminates them again according to the abnormal feature points determined again, and obtains the rock texture point cloud data after being eliminated again. Raindrops or other abnormal feature points may be noise caused by rain, moisture or other environmental factors. These abnormal points will have a negative impact on the quality of the detected rock texture point cloud data. Eliminating these abnormal points again can effectively improve the purity and accuracy of the point cloud data;

[0086] The realization principle of whether there are abnormal feature points of raindrops in the rock texture point cloud data after elimination again:

[0087] First, collect the distance difference data d (p,q) and intensity difference data Δn (p,q) And make a second judgment to obtain the rock texture point cloud data Zpd (p,q) , the specific algorithm formula is:

[0088]

[0089] Among them, d refers to the set standard distance threshold, and Δn refers to the set standard intensity difference threshold;

[0090] When the distance difference data d (p,q) Less than the set standard distance threshold d, and the intensity difference data Δn (p,q) When the intensity difference is less than the set standard intensity difference threshold Δn, Zpd (p,q) =1, indicating that there are abnormal feature points of rain water flowers in the rock texture point cloud data after elimination;

[0091] When the distance difference data d (p,q) When the distance is greater than or equal to the set standard distance threshold d, and the intensity difference data Δn (p,q) When it is greater than or equal to the set standard intensity difference threshold Δn, Pd ng =0, indicating that there are no abnormal feature points of raindrops in the rock texture point cloud data after elimination again.

[0092] The sorting and selecting unit 3 is used to receive the rock texture point cloud data after the re-elimination from the data judgment module 22 and the historical data of the detection and modeling module 11, obtain the laser echo intensity corresponding to the rock texture point cloud data after the re-elimination from the historical data, and sort them in order from low to high according to the laser echo intensity, select the rock texture point cloud data after the re-elimination with the highest laser echo intensity, and transfer the selected rock texture point cloud data to the hidden danger point marking module 12;

[0093] The marking hidden danger point module 12 receives the rock texture point cloud data selected in the sorting and selecting unit 3, inputs the selected rock texture point cloud data and the corresponding three-dimensional coordinates into the three-dimensional model data of the hillside terrain, and the three-dimensional model of the hillside terrain compares the selected rock texture point cloud data of the same three-dimensional coordinate point with the collected rock texture image data to compare whether there is a texture difference, and identifies the three-dimensional coordinate point as a geological disaster hidden danger point according to the comparison result. The comparison result can reveal the changes in the rock surface and help to find the geological disaster hidden danger point early, such as landslide and collapse, which helps to take measures in advance to reduce the risk of geological disasters.

[0094] Specific comparison:

[0095] Case 1: When the selected rock texture point cloud data is different from the rock texture in the collected rock texture image data, it means that the rainfall has caused the rocks on the hillside to move and change, and the three-dimensional coordinate point corresponding to the selected rock texture point cloud data is a geological disaster risk point;

[0096] Case 2: When the selected rock texture point cloud data is the same as the rock texture in the collected rock texture image data, it means that the rainfall has not caused the rocks on the hillside to move or change, and the three-dimensional coordinate point corresponding to the selected rock texture point cloud data is not a geological disaster risk point.

[0097] The second object of the present invention is to provide a method for operating the above-mentioned image data mutual feedback type geological disaster detection system, comprising the following method steps:

[0098] S1, the detection and modeling module 11 uses a drone equipped with a laser scanner to detect the hillside terrain and obtain point cloud data, collects images of rock textures based on the detected point cloud data, and records the corresponding three-dimensional coordinates, and generates a three-dimensional model of the hillside terrain based on the detected point cloud data;

[0099] S2, the hidden danger point marking module 12 marks the three-dimensional model of the hillside terrain according to the corresponding three-dimensional coordinates, and inputs the collected rock texture image data into the three-dimensional model of the marked location;

[0100] S3, the rock detection module 21 uses the drone to accurately locate the rock position according to the three-dimensional coordinates fed back by the three-dimensional model of the hillside terrain, and the drone then uses the laser scanner equipped to detect the rock texture and obtain the point cloud data of the rock texture;

[0101] S4, the data judgment module 22 filters the rock texture point cloud data according to the detected rock texture point cloud data and the number of rock detections, and then judges whether the detected rock texture point cloud data has unclear rock texture and removes it according to the laser echo intensity in the historical data and the point cloud density in the historical data;

[0102] S5, the detection modeling module 11 receives the rock texture point cloud data after being eliminated by the data judgment module 22, and calculates the distance difference and the intensity difference, and then uses the distance difference data and the intensity difference data to make another judgment, and again judges whether there are abnormal feature points of rain water flowers in the rock texture point cloud data after being eliminated and eliminates them again;

[0103] S6. The sorting and selection unit 3 is used to receive the rock texture point cloud data after the re-elimination from the data judgment module 22 and sort them in sequence from low to high, and input the selected rock texture point cloud data and the corresponding three-dimensional coordinates into the three-dimensional model data of the hillside terrain. The three-dimensional model of the hillside terrain uses the selected rock texture point cloud data of the same three-dimensional coordinate point to compare with the collected rock texture image data to compare whether there is a difference in texture, and identifies the three-dimensional coordinate point as a geological disaster risk point according to the comparison result.

[0104] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. The geological disaster detection system based on image data mutual feedback is characterized by: It comprises a detection marking unit (1), a detection judgment unit (2), and a sorting and selecting unit (3); The detection and marking unit (1) detects the hillside terrain using a drone, generates a three-dimensional model, and collects rock texture image data based on the point cloud data, and inputs the rock texture image data into the three-dimensional model; The detection and judgment unit (2) is used to receive data from the detection and marking unit (1). The hillside terrain model feeds back three-dimensional coordinates to the drone. The drone detects rock textures based on the three-dimensional coordinates fed back by the three-dimensional model of the hillside terrain to obtain point cloud data of the rock texture. Based on the detected rock texture point cloud data, it is determined whether there is any unclear rock texture and the rock texture is removed. The detection marking unit (1) receives the rock texture point cloud data after being eliminated by the detection judgment unit (2) and calculates the distance difference and the intensity difference, and uses the calculated distance difference and intensity difference data to make a second judgment, and again judges whether the rock texture point cloud data after being eliminated has raindrop abnormal feature points and eliminates them again; The sorting and selecting unit (3) is used to receive the rock texture point cloud data after re-elimination from the detection and judgment unit (2) and the historical data of the detection and marking unit (1), and to sort the rock texture point cloud data after re-elimination from low to high according to the laser echo intensity corresponding to the rock texture point cloud data after re-elimination, select the rock texture point cloud data after elimination with the highest laser echo intensity, and transmit the selected rock texture point cloud data to the detection and marking unit (1). The detection and marking unit (1) compares the selected rock texture point cloud data of the same three-dimensional coordinate point with the collected rock texture image data to compare whether there is a texture difference, and identifies the three-dimensional coordinate point as a geological disaster risk point according to the comparison result.

2. The image data mutual feedback type geological disaster detection system according to claim 1 is characterized in that: The detection and marking unit (1) comprises a detection modeling module (11) and a potential hazard point marking module (12); The detection and modeling module (11) is used to obtain the rainfall time and historical data of the hillside in real time. 15 minutes before the rainfall on the hillside, an unmanned aerial vehicle equipped with a laser scanner is used to detect the hillside terrain and obtain point cloud data. The point cloud data contains the details of the rock texture and three-dimensional coordinate information. The rock texture image is collected based on the detected point cloud data, and the corresponding three-dimensional coordinates are recorded. At the same time, the detected point cloud data is densely interpolated and a three-dimensional model of the hillside terrain is generated. The historical data include point cloud density and laser echo intensity corresponding to rock texture point cloud data; The hidden danger point marking module (12) is used to receive the rock texture image data, corresponding three-dimensional coordinates, and the three-dimensional model of the hillside terrain collected in the detection modeling module (11), mark the three-dimensional model of the hillside terrain according to the corresponding three-dimensional coordinates, and simultaneously input the collected rock texture image data into the three-dimensional model at the marked location.

3. The image data mutual feedback type geological disaster detection system according to claim 2 is characterized in that: The detection and judgment unit (2) comprises a rock detection module (21) and a data judgment module (22); The rock detection module (21) is used to receive the rainfall time of the hillside, the collected rock texture image data, the corresponding three-dimensional coordinates, and the three-dimensional model of the hillside terrain in the detection and modeling module (11). 15 minutes after the hillside rains and when the surrounding area begins to rain, the three-dimensional model of the hillside terrain feeds back the marked three-dimensional coordinates to the drone. The drone accurately locates the rock position according to the marked three-dimensional coordinates. The drone then uses a laser scanner to detect the rock texture and obtain point cloud data of the rock texture. The data judgment module (22) is used to receive historical data from the detection and modeling module (11), receive rock texture point cloud data detected by the rock detection module (21) and the number of times the rock is detected, filter the rock texture point cloud data using a time window filtering method, and then judge whether the detected rock texture point cloud data has unclear rock texture based on the laser echo intensity in the historical data and the point cloud density in the historical data, remove the detected rock texture point cloud data based on the judged unclear rock texture data, and feed the removed rock texture point cloud data back to the three-dimensional model of the hillside terrain in the detection and modeling module (11).

4. The image data mutual feedback type geological disaster detection system according to claim 3 is characterized in that: The implementation principle of judging whether the rock texture is unclear in the data judging module (22) is as follows: First, collect the laser echo intensity I j and point cloud density ρ j And make a judgment to determine the detected rock texture point cloud data p j Whether there is unclear rock texture, and obtain the unclear rock texture data Pd ng , the specific algorithm formula is: Among them, I yz refers to the set standard intensity threshold, ρ yz′ Refers to the set standard density threshold, p j Refers to the jth rock texture point cloud data detected, I j refers to the laser echo intensity corresponding to the j-th rock texture point cloud data, ρ j Refers to the point cloud density corresponding to the j-th rock texture point cloud data; When the laser echo intensity I j Less than the set standard intensity threshold I yz , or the point cloud density ρ j Less than the set standard density threshold ρ yz′ When Pd ng =1, indicating that the detected rock texture point cloud data p j There are unclear rock textures; When the laser echo intensity I j Greater than or equal to the set standard intensity threshold I yz When, or the point cloud density ρ j Greater than or equal to the set standard density threshold ρ yz′ When Pd ng =0, indicating that the detected rock texture point cloud data p j There is no unclear rock texture.

5. The image data mutual feedback type geological disaster detection system according to claim 3 is characterized in that: The detection modeling module (11) receives the rock texture point cloud data removed by the data judgment module (22), analyzes the point cloud data adjacent to the removed rock texture point cloud data based on the three-dimensional model data of the hillside terrain, calculates the distance difference between the removed rock texture point cloud data and the adjacent point cloud data, then takes the laser echo intensity corresponding to the removed rock texture point cloud data and the adjacent point cloud data from the historical data, calculates the intensity difference based on the laser echo intensity, and feeds the distance difference and intensity difference data back to the data judgment module (22).

6. The image data mutual feedback type geological disaster detection system according to claim 5 is characterized in that: The data judgment module (22) receives the distance difference and intensity difference data from the detection modeling module (11), and uses the distance difference data and the intensity difference data to perform a second judgment to determine whether the rock texture point cloud data after elimination contains abnormal feature points of raindrops, and eliminates the abnormal feature points again based on the abnormal feature points determined again, thereby obtaining the rock texture point cloud data after elimination again.

7. The image data mutual feedback type geological disaster detection system according to claim 6 is characterized in that: The realization principle of whether the rock texture point cloud data after being eliminated again in the data judgment module (22) contains abnormal characteristic points of raindrops: First, collect the distance difference data d (p,q) and intensity difference data Δn (p,q) And make a second judgment to obtain the rock texture point cloud data Zpd (p,q) , the specific algorithm formula is: Among them, d refers to the set standard distance threshold, and Δn refers to the set standard intensity difference threshold; When the distance difference data d (p,w) Less than the set standard distance threshold d, and the intensity difference data Δn (p,q) When the intensity difference is less than the set standard intensity difference threshold Δn, Zpd (p,q) =1, indicating that there are abnormal feature points of raindrops in the rock texture point cloud data after elimination; When the distance difference data d (p,q) When the distance is greater than or equal to the set standard distance threshold d, and the intensity difference data Δn (p,q) When it is greater than or equal to the set standard intensity difference threshold Δn, Pd ng =0, indicating that there are no abnormal feature points of raindrops in the rock texture point cloud data after elimination again.

8. The image data mutual feedback type geological disaster detection system according to claim 6 is characterized in that: The sorting and selecting unit (3) is used to receive the rock texture point cloud data after re-elimination from the data judgment module (22) and the historical data of the detection and modeling module (11), obtain the laser echo intensity corresponding to the rock texture point cloud data after re-elimination from the historical data, and sort them in order from low to high according to the laser echo intensity, select the rock texture point cloud data after re-elimination with the highest laser echo intensity, and transmit the selected rock texture point cloud data to the hidden danger point marking module (12).

9. The image data mutual feedback type geological disaster detection system according to claim 8 is characterized in that: The hidden danger point marking module (12) receives the rock texture point cloud data selected in the sorting and selecting unit (3), inputs the selected rock texture point cloud data and the corresponding three-dimensional coordinates into the three-dimensional model data of the hillside terrain, compares the selected rock texture point cloud data of the same three-dimensional coordinate point with the collected rock texture image data in the three-dimensional model of the hillside terrain, compares whether there is a difference in texture, and identifies the three-dimensional coordinate point as a geological disaster hidden danger point according to the comparison result; Specific comparison: Case 1: When the selected rock texture point cloud data is different from the rock texture in the collected rock texture image data, it means that the rainfall has caused the rocks on the hillside to move and change, and the three-dimensional coordinate point corresponding to the selected rock texture point cloud data is a geological disaster risk point; Case 2: When the selected rock texture point cloud data is the same as the rock texture in the collected rock texture image data, it means that the rainfall has not caused the rocks on the hillside to move or change, and the three-dimensional coordinate point corresponding to the selected rock texture point cloud data is not a geological disaster risk point.

10. A method for operating the image data mutual feedback type geological disaster detection system according to any one of claims 1 to 9, characterized in that: The method comprises the following steps: S1, the detection and modeling module (11) uses the unmanned aerial vehicle equipped with a laser scanner to detect the hillside terrain and obtain point cloud data, collects images of rock textures based on the detected point cloud data, and records the corresponding three-dimensional coordinates, and generates a three-dimensional model of the hillside terrain based on the detected point cloud data; S2, a module for marking hidden danger points (12) marks the three-dimensional model of the hillside terrain according to the corresponding three-dimensional coordinates, and inputs the collected rock texture image data into the three-dimensional model of the marked location; S3, rock detection module (21) uses the drone to accurately locate the rock position according to the three-dimensional coordinates fed back by the three-dimensional model of the hillside terrain, and the drone then uses the laser scanner equipped to detect the rock texture and obtain point cloud data of the rock texture; S4, the data judgment module (22) filters the rock texture point cloud data according to the detected rock texture point cloud data and the number of times the rock is detected, and then judges whether the detected rock texture point cloud data has unclear rock texture and removes it according to the laser echo intensity in the historical data and the point cloud density in the historical data; S5, the detection modeling module (11) receives the rock texture point cloud data after being eliminated by the data judgment module (22), calculates the distance difference and the intensity difference, and then uses the distance difference data and the intensity difference data to make another judgment, and again judges whether there are raindrop abnormal feature points in the rock texture point cloud data after being eliminated, and then eliminates them again; S6, the sorting and selecting unit (3) is used to receive the rock texture point cloud data after the re-elimination from the data judgment module (22), sort them in order from low to high, input the selected rock texture point cloud data and the corresponding three-dimensional coordinates into the three-dimensional model data of the hillside terrain, and compare the selected rock texture point cloud data of the same three-dimensional coordinate point with the collected rock texture image data in the three-dimensional model of the hillside terrain to compare whether there is a texture difference, and identify the three-dimensional coordinate point as a geological disaster risk point according to the comparison result.

Images