A method for rapid comparative identification of slope deformation without a target

Abstract

The present invention relates to a method for quickly comparing and identifying slope deformation without a target, which relates to the field of slope deformation monitoring, and includes: identifying the similarity of the first and second slope images to obtain the slope similarity. When the similarity is greater than the threshold, performing saliency analysis on the first and second slope images to obtain the first and second slope saliency maps; respectively performing screening of the coordinates of the significant regions to obtain the first and second significant region coordinate distributions, and obtaining the basic deformation coefficient through comparative analysis; analyzing and obtaining the moving error coefficient of the mobile monitoring device, and combining the slope similarity to perform error compensation on the basic deformation coefficient to obtain the compensated deformation coefficient interval as the deformation identification result. Through the present application, the technical problems that the traditional method relies on targets and fixed monitoring instruments, has high installation and maintenance costs, cannot flexibly adapt to environmental changes, and has low accuracy and efficiency in deformation identification can be solved; rapid, low-cost and high-precision slope deformation identification can be achieved.

Description

Technical Field

[0001] The present invention relates to the field of slope deformation monitoring, and particularly to a method for quickly comparing and identifying slope deformation without a target. Background Art

[0002] Slope deformation monitoring is an important link in geological disaster prevention and control, especially of great significance for the early warning of geological disasters such as landslides and collapses.

[0003] Currently, traditional slope deformation monitoring methods mainly rely on targets and fixed monitoring instruments, such as total stations, GPS, inclinometers and other devices. These devices judge the potential risks of geological disasters by accurately measuring the displacement and angle changes of the targets or directly monitoring the deformation of the slope surface.

[0004] However, traditional slope deformation monitoring methods have multiple technical problems and deficiencies, such as high cost and high maintenance requirements, poor environmental adaptability, low accuracy and efficiency, etc., which restrict the accuracy, efficiency and application scope of slope deformation monitoring. Summary of the Invention

[0005] In view of the technical problems that the traditional slope deformation monitoring method relies on targets and fixed monitoring instruments, has high installation and maintenance costs, and cannot flexibly adapt to environmental changes, with low accuracy and efficiency in deformation identification, the present invention provides a method for quickly comparing and identifying slope deformation without a target to solve this problem.

[0006] The technical solution of the present invention to solve the above technical problems is as follows:

[0007] In a first aspect, the present invention provides a method for quickly comparing and identifying slope deformation without a target, including: by means of a mobile monitoring device, at preset moments within a first deformation monitoring period and a second deformation monitoring period, collecting a first slope image and a second slope image of a target slope at a preset azimuth, wherein the mobile monitoring device includes a drone or a mobile vehicle; performing similarity recognition on the first slope image and the second slope image to obtain a slope similarity, and when the slope similarity is greater than a similarity threshold, performing saliency analysis on the first slope image and the second slope image to obtain a first slope saliency map and a second slope saliency map; respectively performing screening on the coordinates of the salient regions of the first slope saliency map and the second slope saliency map to obtain a first salient region coordinate distribution and a second salient region coordinate distribution, and through comparative analysis, obtaining a basic deformation coefficient of the target slope; based on the historical movement data of the mobile monitoring device, analyzing to obtain a movement error coefficient, and combining with the slope similarity, performing error compensation on the basic deformation coefficient to obtain a compensated deformation coefficient interval as the slope deformation identification result.

[0008] In a second aspect, the present invention provides a targetless rapid comparison and identification system for slope deformation, comprising: a slope image acquisition module, configured to collect a first slope image and a second slope image of a target slope at a preset time and in a preset direction within a first deformation monitoring period and a second deformation monitoring period through a mobile monitoring device, wherein the mobile monitoring device includes a drone or a mobile vehicle; an image saliency analysis module, configured to perform similarity recognition on the first slope image and the second slope image to obtain a slope similarity, and when the slope similarity is greater than a similarity threshold, perform saliency analysis on the first slope image and the second slope image to obtain a first slope saliency map and a second slope saliency map; a basic deformation coefficient analysis module, configured to respectively perform significant region coordinate screening on the first slope saliency map and the second slope saliency map to obtain a first significant region coordinate distribution and a second significant region coordinate distribution, and perform comparative analysis to obtain the basic deformation coefficient of the target slope; a basic deformation coefficient compensation module, configured to analyze and obtain a movement error coefficient based on the historical movement data of the mobile monitoring device, and combine the slope similarity to perform error compensation on the basic deformation coefficient to obtain a compensated deformation coefficient interval as the slope deformation identification result.

[0009] The beneficial effects of the present invention are as follows: By means of a mobile monitoring device, a first slope image and a second slope image of a target slope are collected at a preset time and in a preset direction within a first deformation monitoring period and a second deformation monitoring period, wherein the mobile monitoring device includes a drone or a mobile vehicle; then, similarity recognition is performed on the first slope image and the second slope image to obtain a slope similarity, and when the slope similarity is greater than a similarity threshold, saliency analysis is performed on the first slope image and the second slope image to obtain a first slope saliency map and a second slope saliency map; then, significant region coordinate screening is respectively performed on the first slope saliency map and the second slope saliency map to obtain a first significant region coordinate distribution and a second significant region coordinate distribution, and comparative analysis is performed to obtain the basic deformation coefficient of the target slope; further, based on the historical movement data of the mobile monitoring device, a movement error coefficient is analyzed and obtained, and in combination with the slope similarity, error compensation is performed on the basic deformation coefficient to obtain a compensated deformation coefficient interval, and the compensated deformation coefficient interval is used as the slope deformation identification result; that is to say, by flexibly collecting images through a drone or a mobile vehicle, combining technologies such as similarity analysis, saliency analysis, and error compensation, targetless slope deformation monitoring can be achieved, avoiding the need for high-cost target installation and maintenance, and at the same time, being able to flexibly respond to complex terrain and environmental changes, realizing rapid, low-cost, and high-precision slope deformation identification, achieving the technical effects of significantly improving the accuracy, reliability, and efficiency of deformation identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1Schematic flow chart of a method for rapid comparison and identification of slope deformation without a target provided by the present invention;

[0011] Figure 2 Schematic structural diagram of a system for rapid comparison and identification of slope deformation without a target provided by the present invention.

[0012] In the accompanying drawings, the components represented by each reference numeral are described as follows:

[0013] Slope image acquisition module 01, image saliency analysis module 02, basic deformation coefficient analysis module 03, basic deformation coefficient compensation module 04. Detailed implementation manners

[0014] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of the present invention.

[0015] In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the described features. In the description of the present invention, "a plurality of" means two or more, unless otherwise specifically defined.

[0016] In the description of the present invention, the term "for example" is used to mean "serving as an example, illustration, or explanation". Any embodiment described as "for example" in the present invention is not necessarily construed as being more preferred or advantageous than other embodiments. In order for any person skilled in the art to implement and use the present invention, the following description is given. In the following description, details are set forth for the purpose of explanation. It should be understood that those skilled in the art can recognize that the present invention can be implemented without these specific details. In other instances, well-known structures and processes are not elaborated in detail to avoid unnecessary details from obscuring the description of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope that conforms to the principles and features disclosed in the present invention.

[0017] Embodiment 1, as Figure 1 shown, the embodiment of the present invention provides a method for rapid comparison and identification of slope deformation without a target, which specifically includes the following steps:

[0018] S100: At preset times within the first deformation monitoring period and the second deformation monitoring period, use a mobile monitoring device to collect a first slope image and a second slope image of the target slope in a preset direction, where the mobile monitoring device includes a drone or a mobile vehicle.

[0019] Furthermore, step S100 of the present invention further includes:

[0020] S110: Use a mobile monitoring device to respectively control it to move to a preset direction at preset times within the first deformation monitoring period and the second deformation monitoring period, where the first deformation monitoring period and the second deformation monitoring period are two adjacent front and rear deformation monitoring periods; S120: At the preset direction, respectively collect and obtain a first slope image and a second slope image.

[0021] Specifically, first, select a mobile monitoring device, which includes a drone or a mobile vehicle. By using the drone or the mobile vehicle as the mobile monitoring device, its flexibility enables the monitoring personnel to avoid relying on fixed monitoring points and can perform dynamic monitoring of the slope from multiple angles and positions. At the same time, the device can adjust the flight or driving path according to the monitoring requirements to ensure the accuracy of the acquisition range and angle of the image data. Then, use the mobile monitoring device to respectively control it to move to a preset direction at preset times within the first deformation monitoring period and the second deformation monitoring period, where the first deformation monitoring period and the second deformation monitoring period are two adjacent front and rear deformation monitoring periods, and the deformation monitoring period can be set according to actual monitoring requirements. For example, if the deformation monitoring period is set to 1 day and the monitoring time is 2 pm every day, then the monitoring time of the first deformation monitoring period is 2 pm of the previous day, and the monitoring time of the second deformation monitoring period is 2 pm of the next day. The preset direction includes the position (fixed acquisition position) and angle (fixed acquisition angle) of image acquisition, which can be set according to the actual monitoring scenario to ensure the consistency of the acquired images in space and angle.

[0022] Then, at the preset direction, use the mobile monitoring device (drone or mobile vehicle) to collect images of the slope to obtain the first slope image of the first deformation monitoring period and the second slope image of the second deformation monitoring period. By obtaining the first slope image and the second slope image, it provides basic data for subsequent slope deformation analysis.

[0023] S200: Perform similarity recognition on the first slope image and the second slope image to obtain the slope similarity. When the slope similarity is greater than the similarity threshold, perform saliency analysis on the first slope image and the second slope image to obtain a first slope saliency map and a second slope saliency map.

[0024] Furthermore, step S200 of the present invention further includes:

[0025] S210: Obtain a pre-trained slope similarity recognition channel, where the slope similarity recognition channel is obtained by collecting multiple groups of sample slope images and the marked sample slope similarity set and training to convergence based on a siamese network; S220: Input the combination of the first slope image and the second slope image into the slope similarity recognition channel, and recognize and output to obtain the slope similarity.

[0026] Specifically, first, construct a slope similarity recognition channel based on a siamese network. A siamese network is a special neural network architecture composed of two identical sub-networks (usually sharing the same weights), which is used to process a pair of input data. Each sub-network independently extracts the features of the input data and maps these features into the same feature space, and then calculates the similarity between the two through a metric function (such as Euclidean distance or cosine similarity). In the slope similarity recognition task, the input of the siamese network is two slope images (for example, the slope image in the first monitoring period and the slope image in the second monitoring period), and these two images will be processed through the two sub-networks of the network respectively, and the similarity measurement value is output; the slope similarity recognition channel includes an input layer, a similarity comparison layer, and an output layer, where the input data of the output layer is two slope images, and the output data of the output layer is the slope similarity between the two slope images.

[0027] Next, collect multiple groups of sample slope images, and mark the slope similarity for each group of sample slope images, that is, label each group of images with a similarity label, indicating the similarity between the two images. Numerical labels (such as similarity values from 0 to 1) can be used to represent the similarity between image pairs, and a sample slope similarity set is obtained. Further, using each group of sample slope images as the input and the sample slope similarity as the output, use the multiple groups of sample slope images and the sample slope similarity set as training data to perform supervised training on the slope similarity recognition channel. First, the input image pair is processed through two sub-networks to generate the feature vectors of each pair of images and calculate the similarity between them; then, calculate the difference between the predicted value of the model and the true label through the loss function to obtain the loss value; then, calculate the gradient of each weight parameter through the backpropagation algorithm, update the network parameters, and reduce the loss. Backpropagation will adjust the weights of the convolutional layer and the fully connected layer in the sub-network, enabling the network to better extract features helpful for judging similarity; further, use an optimization algorithm (such as Adam or SGD) to adjust the network parameters according to the gradient to minimize the loss function. Each round of training will update the model parameters through the gradient descent algorithm, making the performance of the model on the training set better and better; perform iterative training until the loss function converges to obtain the trained slope similarity recognition channel. By constructing a slope similarity recognition channel based on a siamese network, the similarity between the input slope image groups can be predicted, significantly improving the scientificity, accuracy, and efficiency of image similarity analysis.

[0028] Then, the first slope image and the second slope image are combined and input into the slope similarity recognition channel for image similarity recognition, and the slope similarity is output. Among them, the greater the slope similarity, the more similar the morphologies of the slope images collected twice are, indicating that the slope deformation is small or there is no deformation. On the contrary, the smaller the slope similarity, the greater the slope deformation, and there may be risks of geological disasters such as landslides and collapses.

[0029] S230: Determine whether the slope similarity is greater than the slope similarity threshold. If so, perform saliency analysis on the first slope image and the second slope image to obtain a first slope saliency map and a second slope saliency map.

[0030] Furthermore, step S230 of the present invention further includes:

[0031] S231: Perform grayscale processing on the first slope image and the second slope image to obtain a first grayscale slope image and a second grayscale slope image; S232: Extract first slope pixel points in the first grayscale slope image, and randomly extract multiple neighboring slope gray values of multiple neighboring slope pixel points adjacent to the first slope pixel points; S233: Randomly extract multiple random slope gray values of multiple random slope pixel points in the first grayscale slope image, calculate the difference ratio between the mean value of the multiple neighboring slope gray values and the mean value of the multiple random slope gray values to obtain a first saliency; S234: Continue to calculate the saliency of all slope pixel points in the first grayscale slope image to obtain a first slope saliency map; S235: Perform saliency analysis on the second slope image to obtain a second slope saliency map.

[0032] Specifically, determine whether the slope similarity is greater than the slope similarity threshold. The slope similarity threshold is usually determined based on previous experience or experiments, and its setting depends on the conditions of image acquisition, the expected degree of slope deformation, and the characteristics of the similarity metric used. If the slope similarity is greater than the slope similarity threshold, indicating that the similarity gap between the two is small, then perform saliency analysis on the first slope image and the second slope image.

[0033] First, perform grayscale processing on the first slope image and the second slope image. Grayscale processing is the process of converting a color image into a grayscale image. The purpose is to simplify the computational complexity in image processing while retaining the important structures and texture information in the image. In slope deformation monitoring, the grayscale image can highlight the brightness changes in the deformation area, making the subsequent saliency analysis processing more efficient and accurate. Commonly used grayscale processing methods include the weighted average method, etc.; obtain a first grayscale slope image and a second grayscale slope image.

[0034] Next, randomly select any pixel point in the first grayscale slope image as the first slope pixel point, and randomly extract the grayscale values of multiple adjacent slope pixel points adjacent to the first slope pixel point, that is, around the selected first slope pixel point, with this point as the center, randomly extract multiple adjacent slope pixel points. The coordinates of these adjacent pixel points should be within the neighborhood of the first slope pixel point. A neighborhood range can be set (such as a window size of 3*3 or 5*5. Setting a smaller neighborhood such as a 3*3 window can improve the accuracy of saliency analysis calculation), so as to ensure that the selected adjacent pixel points are close enough to the central pixel point. Then randomly extract the grayscale values of multiple random slope pixel points in the first grayscale slope image, where the number of random slope pixel points is the same as the number of adjacent slope pixel points. Further calculate the mean value of the multiple adjacent slope grayscale values to obtain the adjacent slope grayscale mean value, and calculate the mean value of the multiple random slope grayscale values to obtain the random slope grayscale mean value; then calculate the difference ratio between the adjacent slope grayscale mean value and the random slope grayscale mean value, where the difference ratio is the ratio of the difference between the adjacent slope grayscale mean value and the random slope grayscale mean value to the random slope grayscale mean value. For example, if the adjacent slope grayscale mean value is 110 and the random slope grayscale mean value is 100, then the deviation ratio is (110 - 100) / 100 = 0.1; and set the difference ratio as the first saliency.

[0035] Then, using the same method for calculating the first saliency, continue to calculate the saliency of all slope pixel points in the first grayscale slope image, and construct the first slope saliency map in combination with the position coordinates of the pixel points. Further, use the same method to perform saliency analysis on the second slope image to generate the second slope saliency map.

[0036] S240: If not, then re-collect the slope image. When the slope similarity of a preset number of continuously collected slope images is not greater than the slope similarity threshold, output a slope deformation prompt message for prompting.

[0037] Specifically, if the slope similarity is less than or equal to the slope similarity threshold, it indicates that the difference between the images is relatively large, which may be caused by external environmental factors (such as weather changes, lighting problems, equipment errors, etc.). In this case, the similarity of the image group is not high enough, and it may not be possible to accurately compare the slope deformations. Then, start the continuous acquisition mechanism, re-acquire a new second slope image and calculate its similarity with the first slope image; and when the slope similarities of the preset number of (which can be set according to the actual scenario, such as continuously acquiring 5 times) slope images acquired continuously are all not greater than the slope similarity threshold, that is, when the similarities of the continuously acquired slope images are all lower than the set threshold, it can be judged that this is not an error caused by external environment (such as weather, lighting, etc. changes), but a real slope deformation signal. At this time, output a slope deformation prompt message for prompting.

[0038] S300: Screen the significant region coordinates of the first slope saliency map and the second slope saliency map respectively to obtain the first significant region coordinate distribution and the second significant region coordinate distribution, and obtain the basic deformation coefficient of the target slope through comparative analysis.

[0039] Furthermore, step S300 of the present invention further includes:

[0040] S310: Screen the slope pixel points in the first slope saliency map with saliency greater than the saliency threshold to obtain a plurality of first significant slope pixel points; S320: Cluster the adjacent first significant slope pixel points with the adjacent aggregation quantity greater than the aggregation quantity threshold into the first significant region to obtain a plurality of first significant regions, and combine the coordinates of the first significant slope pixel points in the plurality of first significant regions to construct and obtain the first significant region coordinate distribution; S330: Screen the significant region coordinates of the second slope saliency map to obtain the second significant region coordinate distribution.

[0041] Specifically, first, set a suitable saliency threshold according to the actual situation, which can be set after analyzing historical sample data; then screen the saliencies of a plurality of pixel points in the first slope saliency map according to the saliency threshold, and set the slope pixel points greater than the saliency threshold as the first significant slope pixel points to obtain a plurality of first significant slope pixel points.

[0042] Next, set a clustering quantity threshold (for example, set the quantity threshold to 50); and cluster adjacent first significant slope pixel points with adjacent clustering quantities greater than the clustering quantity threshold into a first significant region. That is, for each significant pixel point and its adjacent pixel points, if they also meet the significance threshold, they are considered adjacent and can be clustered into the same significant region, obtaining multiple first significant regions. Then, combine the coordinates of the first significant slope pixel points within the multiple first significant regions to construct a coordinate distribution of the first significant regions. Then, based on the significance threshold and the clustering quantity threshold, perform significant region coordinate screening on the second slope significance map to obtain multiple second significant regions, and construct a coordinate distribution of the second significant regions.

[0043] S340: Conduct a comparative analysis on the coordinate distribution of the first significant regions and the coordinate distribution of the second significant regions to obtain the basic deformation coefficient of the target slope.

[0044] Furthermore, step S340 of the present invention further includes:

[0045] S341: Traverse and calculate the repetition ratio of the coordinates of multiple first significant regions within the coordinate distribution of the first significant regions and the coordinates of multiple second significant regions within the coordinate distribution of the second significant regions, screen the largest multiple repetition ratios, and obtain multiple significant region coordinate repetition ratios; S342: Calculate the mean value of the multiple significant region coordinate repetition ratios to obtain the average repetition ratio; S343: Subtract the average repetition ratio from 1 to obtain the basic deformation coefficient.

[0046] Specifically, first, traverse and calculate the repetition ratio of the coordinates of multiple first significant regions within the coordinate distribution of the first significant regions and the coordinates of multiple second significant regions within the coordinate distribution of the second significant regions. For example, calculate the overlapping pixels of the two regions through spatial distance or overlapping area to evaluate the coordinate coincidence degree. Among them, the repetition ratio represents the coincidence degree of the two significant regions in spatial coordinates. The higher the repetition ratio, the smaller the change between the two regions. On the contrary, it indicates the existence of deformation or difference, obtaining multiple region repetition ratios; then, screen the largest multiple repetition ratios (for example, select the largest 50% of the repetition ratios) in the multiple region repetition ratios and set them as the significant region coordinate repetition ratios, obtaining multiple significant region coordinate repetition ratios.

[0047] Then, calculate the mean value of the repetition ratios of the coordinates of the multiple significant regions to obtain the average repetition ratio. Next, subtract the average repetition ratio from 1, and use the difference between the two as the basic deformation coefficient. Among them, the larger the average repetition ratio, the less obvious deformation is indicated, and the basic deformation coefficient is close to 0. The smaller the average repetition ratio, the larger the value of the basic deformation coefficient, indicating that a larger deformation has occurred. By calculating and analyzing the repetition ratios of the coordinates of the significant regions, the basic deformation coefficient can be obtained, reducing the influence of environmental interference factors (such as lighting, weather, etc.), so as to obtain more stable and accurate deformation analysis results under different environmental conditions, improving the accuracy and reliability of slope deformation analysis.

[0048] S400: Based on the historical movement data of the mobile monitoring device, analyze and obtain the movement error coefficient, and combine it with the slope similarity to perform error compensation on the basic deformation coefficient to obtain the compensated deformation coefficient interval as the slope deformation identification result.

[0049] Furthermore, step S400 of the present invention further includes:

[0050] S410: Obtain the actual movement azimuth parameters of the mobile monitoring device under the same azimuth movement instruction within the historical time to obtain the historical actual movement parameter set; S420: Calculate the average azimuth deviation between the historical actual movement parameter set and the same azimuth movement instruction to obtain the movement error coefficient; S430: Subtract the slope similarity from 1 to obtain the image error coefficient; S440: Use the movement error coefficient and the image error coefficient to perform error compensation on the basic deformation coefficient to obtain the compensated deformation coefficient interval as the slope deformation identification result.

[0051] Specifically, when using a mobile monitoring device (such as a drone or a mobile vehicle) to collect slope images, errors may occur, including equipment position errors (such as offsets), photographing direction errors (such as inconsistent angles), and image acquisition errors (such as unstable image quality). These errors may lead to inaccuracies in slope deformation analysis results. Therefore, it is necessary to eliminate the influence of errors.

[0052] First, obtain the actual movement azimuth parameters of the mobile monitoring device under the same azimuth movement instruction within the historical time at multiple monitoring time nodes. Since there may be errors in the movement of the device, there may be deviations between the actual movement azimuth and the preset azimuth. The actual movement azimuth refers to the azimuth where the device finally actually arrives (including the shooting position and shooting angle) when executing the movement instruction, and a set of historical actual movement parameters is obtained. Then, calculate the average azimuth deviation between the set of historical actual movement parameters and the same azimuth movement instruction, that is, for each movement instruction at a historical time point, calculate the difference between the actual azimuth of the device and the preset azimuth. Usually, the Euclidean distance (position deviation) and the angular difference (direction deviation) can be used to represent it, and multiple azimuth deviations are obtained, and the mean value is calculated to obtain the movement error coefficient.

[0053] On the other hand, subtract the slope similarity from 1, and use the difference between the two as the image error coefficient; then calculate the mean value of the movement error coefficient and the image error coefficient to obtain the comprehensive error coefficient, and perform error compensation on the basic deformation coefficient according to the comprehensive error coefficient to obtain the compensated deformation coefficient interval. For example, assume that the movement error coefficient and the image error coefficient are 0.01 and 0.03 respectively, and the basic deformation coefficient is 0.1, then the comprehensive error coefficient is 0.02, and the compensated deformation coefficient interval is 0.1 ± 0.02. Finally, use the compensated deformation coefficient interval as the slope deformation identification result. By combining the movement error coefficient and the image error coefficient and using the mean value to calculate the comprehensive error coefficient, the error compensation of the basic deformation coefficient for slope deformation monitoring can be performed, and a compensated deformation coefficient interval can be obtained. This interval can provide a range of deformation amplitudes, which can reduce or avoid deviations caused by equipment or environmental errors, and further improve the accuracy and reliability of slope deformation monitoring.

[0054] The method for rapid comparison and identification of slope deformation without a target provided by the embodiment of the present invention has at least the following technical effects:

[0055] By flexibly collecting images with a drone or a mobile vehicle, combining technologies such as similarity analysis, saliency analysis, and error compensation, the slope deformation monitoring without a target can be realized, avoiding the need for high-cost target installation and maintenance, and at the same time, it can flexibly respond to complex terrain and environmental changes, realizing rapid, low-cost, and high-precision slope deformation identification, achieving the technical effects of significantly improving the accuracy, reliability, and efficiency of deformation identification.

[0056] Embodiment 2, as Figure 2 shown, based on the same inventive concept as the method for rapid comparison and identification of slope deformation without a target provided in Embodiment 1, the embodiment of the present invention further provides a system for rapid comparison and identification of slope deformation without a target, including:

[0057] The slope image acquisition module 01 is used to collect the first slope image and the second slope image of the target slope at a preset time within the first deformation monitoring period and the second deformation monitoring period through a mobile monitoring device, where the mobile monitoring device includes a drone or a mobile vehicle; the image saliency analysis module 02 is used to identify the similarity between the first slope image and the second slope image to obtain the slope similarity. When the slope similarity is greater than the similarity threshold, perform saliency analysis on the first slope image and the second slope image to obtain the first slope saliency map and the second slope saliency map; the basic deformation coefficient analysis module 03 is used to respectively screen the significant region coordinates of the first slope saliency map and the second slope saliency map to obtain the first significant region coordinate distribution and the second significant region coordinate distribution, and perform comparative analysis to obtain the basic deformation coefficient of the target slope; the basic deformation coefficient compensation module 04 is used to analyze and obtain the movement error coefficient based on the historical movement data of the mobile monitoring device, and combine the slope similarity to perform error compensation on the basic deformation coefficient to obtain a compensated deformation coefficient interval as the slope deformation identification result.

[0058] Further, the non-targeted slope deformation rapid comparison and identification system further includes: controlling to move to a preset orientation respectively at preset times within the first deformation monitoring period and the second deformation monitoring period through a mobile monitoring device, where the first deformation monitoring period and the second deformation monitoring period are two adjacent front and back deformation monitoring periods; collecting and obtaining the first slope image and the second slope image respectively at the preset orientation.

[0059] Further, the non-targeted slope deformation rapid comparison and identification system further includes: obtaining a pre-trained slope similarity recognition channel, where the slope similarity recognition channel is obtained by collecting multiple groups of sample slope images and a marked sample slope similarity set and training to convergence based on a siamese network; combining and inputting the first slope image and the second slope image into the slope similarity recognition channel, and identifying and outputting to obtain the slope similarity; determining whether the slope similarity is greater than the slope similarity threshold. If so, perform saliency analysis on the first slope image and the second slope image to obtain the first slope saliency map and the second slope saliency map; if not, re-collect slope images, and when the slope similarities of the preset number of continuously collected slope images are not greater than the slope similarity threshold, output a slope deformation prompt message for prompting.

[0060] Further, the non-target slope deformation rapid comparison and identification system further includes: grayscale processing the first slope image and the second slope image to obtain a first grayscale slope image and a second grayscale slope image; extracting first slope pixel points in the first grayscale slope image, and randomly extracting multiple adjacent slope gray values of multiple adjacent slope pixel points adjacent to the first slope pixel points; randomly extracting multiple random slope gray values of multiple random slope pixel points in the first grayscale slope image, calculating the difference ratio between the mean value of the multiple adjacent slope gray values and the mean value of the multiple random slope gray values to obtain a first saliency; continuing to calculate the saliency of all slope pixel points in the first grayscale slope image to obtain a first slope saliency map; performing saliency analysis on the second slope image to obtain a second slope saliency map.

[0061] Further, the non-target slope deformation rapid comparison and identification system further includes: screening slope pixel points in the first slope saliency map with saliency greater than the saliency threshold to obtain multiple first significant slope pixel points; clustering adjacent first significant slope pixel points with an adjacent aggregation quantity greater than the aggregation quantity threshold into a first significant region to obtain multiple first significant regions, and combining the coordinates of the first significant slope pixel points in the multiple first significant regions to construct a first significant region coordinate distribution; performing significant region coordinate screening on the second slope saliency map to obtain a second significant region coordinate distribution; performing comparative analysis on the first significant region coordinate distribution and the second significant region coordinate distribution to obtain the basic deformation coefficient of the target slope.

[0062] Further, the non-target slope deformation rapid comparison and identification system further includes: traversing and calculating the repetition ratio of the coordinates of multiple first significant regions in the first significant region coordinate distribution and the coordinates of multiple second significant regions in the second significant region coordinate distribution, screening the largest multiple repetition ratios to obtain multiple significant region coordinate repetition ratios; calculating the mean value of the multiple significant region coordinate repetition ratios to obtain an average repetition ratio; subtracting the average repetition ratio from 1 to obtain the basic deformation coefficient.

[0063] Further, the non-target slope deformation rapid comparison and identification system further includes: obtaining the actual movement azimuth parameters of the mobile monitoring device under the same azimuth movement instruction in historical time to obtain a historical actual movement parameter set; calculating the average azimuth deviation between the historical actual movement parameter set and the same azimuth movement instruction to obtain a movement error coefficient; subtracting the slope similarity from 1 to obtain an image error coefficient; using the movement error coefficient and the image error coefficient to perform error compensation on the basic deformation coefficient to obtain a compensated deformation coefficient interval as the slope deformation identification result.

[0064] It should be noted that in the above embodiments, the descriptions of the various embodiments have their respective emphases. For parts not described in detail in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0065] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0066] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0067] These computer program instructions can 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, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0068] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0069] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications to these embodiments once they learn the basic creative concepts.

[0070] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalent technologies, the present invention also intends to include these changes and modifications.

Claims

1. A method for rapid comparison and identification of slope deformation without a target, characterized in that: Methods include: By means of a mobile monitoring device, at a preset time in a first deformation monitoring period and a second deformation monitoring period, a first slope image and a second slope image of the target slope are collected at a preset orientation, wherein the mobile monitoring device includes a drone or a mobile vehicle; Performing similarity recognition on the first slope image and the second slope image to obtain slope similarity, and when the slope similarity is greater than a similarity threshold, performing saliency analysis on the first slope image and the second slope image to obtain a first slope saliency map and a second slope saliency map; Screening the slope pixel points whose significance is greater than the significance threshold in the first slope significance map to obtain a plurality of first significant slope pixel points; Clustering adjacent first significant slope pixel points whose adjacent clustering number is greater than a clustering number threshold into a first significant region, obtaining multiple first significant regions, and combining the coordinates of the first significant slope pixel points in the multiple first significant regions to construct a coordinate distribution of the first significant region; Screening the coordinates of significant regions on the second slope significance map to obtain a second significant region coordinate distribution; traversing and calculating the repetition ratios of the coordinates of a plurality of first significant regions in the first significant region coordinate distribution and the coordinates of a plurality of second significant regions in the second significant region coordinate distribution, selecting the largest multiple repetition ratios, and obtaining multiple significant region coordinate repetition ratios; Calculating the mean of the repetition ratios of the plurality of significant region coordinates to obtain an average repetition ratio; The basic deformation coefficient is obtained by subtracting the average repetition ratio from 1; Acquire actual movement azimuth parameters of the mobile monitoring device under the same movement azimuth instruction in historical time, and obtain a historical actual movement parameter set; Calculate the average azimuth deviation between the historical actual movement parameter set and the same azimuth movement instruction to obtain a movement error coefficient; Subtracting the slope similarity from 1 to obtain an image error coefficient; The movement error coefficient and the image error coefficient are used to perform error compensation on the basic deformation coefficient to obtain a compensated deformation coefficient interval as a slope deformation identification result.

2. The method for rapid comparison and identification of slope deformation without a target according to claim 1 is characterized in that: The method collects a first slope image and a second slope image of a target slope at a preset position at a preset time in a first deformation monitoring period and a second deformation monitoring period by a mobile monitoring device, including: By means of a mobile monitoring device, at preset moments in a first deformation monitoring cycle and a second deformation monitoring cycle, the device is respectively controlled to move to a preset position, wherein the first deformation monitoring cycle and the second deformation monitoring cycle are two adjacent deformation monitoring cycles; The first slope image and the second slope image are respectively acquired at the preset positions.

3. The method for rapid comparison and identification of slope deformation without a target according to claim 1 is characterized in that: Performing similarity recognition on the first slope image and the second slope image to obtain slope similarity, and when the slope similarity is greater than a similarity threshold, performing saliency analysis on the first slope image and the second slope image to obtain a first slope saliency map and a second slope saliency map, including: Acquire a pre-trained slope similarity recognition channel, wherein the slope similarity recognition channel is obtained by collecting multiple groups of sample slope images and marked sample slope similarity sets, and training the twin network until convergence; Combining the first slope image and the second slope image into the slope similarity recognition channel, and identifying and outputting the slope similarity; Determine whether the slope similarity is greater than a slope similarity threshold, and if so, perform saliency analysis on the first slope image and the second slope image to obtain a first slope saliency map and a second slope saliency map; If not, the slope image is collected again, and when the slope similarities of a preset number of slope images collected continuously are not greater than the slope similarity threshold, slope deformation prompt information is output for prompting.

4. The method for rapid comparison and identification of slope deformation without a target according to claim 3 is characterized in that: Performing saliency analysis on the first slope image and the second slope image to obtain a first slope saliency map and a second slope saliency map includes: grayscale the first slope image and the second slope image to obtain a first grayscale slope image and a second grayscale slope image; Extracting a first slope pixel point in the first grayscale slope image, and randomly extracting a plurality of adjacent slope grayscale values ​​of a plurality of adjacent slope pixel points adjacent to the first slope pixel point; Randomly extracting a plurality of random slope grayscale values ​​of a plurality of random slope pixel points in the first grayscale slope image, calculating a difference ratio between an average of the plurality of adjacent slope grayscale values ​​and an average of the plurality of random slope grayscale values, and obtaining a first saliency; Continue to calculate and obtain the saliency of all slope pixels in the first grayscale slope image to obtain a first slope saliency map; A saliency analysis is performed on the second side slope image to obtain a second side slope saliency map.

5. A non-target slope deformation rapid comparison and identification system, characterized in that: The steps for implementing the method for rapid comparison and identification of slope deformation without a target as described in any one of claims 1 to 4 include: A slope image acquisition module is used to acquire a first slope image and a second slope image of a target slope at a preset position at a preset time within a first deformation monitoring period and a second deformation monitoring period by a mobile monitoring device, wherein the mobile monitoring device includes a drone or a mobile vehicle; An image saliency analysis module is used to perform similarity recognition on the first slope image and the second slope image to obtain slope similarity, and when the slope similarity is greater than a similarity threshold, perform saliency analysis on the first slope image and the second slope image to obtain a first slope saliency map and a second slope saliency map; A basic deformation coefficient analysis module is used to screen the coordinates of the significant regions of the first slope significance map and the second slope significance map respectively, obtain the coordinate distribution of the first significant region and the coordinate distribution of the second significant region, and compare and analyze to obtain the basic deformation coefficient of the target slope; The basic deformation coefficient compensation module is used to analyze and obtain the movement error coefficient based on the historical movement data of the mobile monitoring device, and to perform error compensation on the basic deformation coefficient in combination with the slope similarity to obtain the compensated deformation coefficient interval as the slope deformation identification result.

Images