Landslide early warning method based on polymorphic target tracking intelligent identification

Abstract

The invention discloses a landslide early warning method based on polymorphic target tracking intelligent identification. The landslide early warning method comprises the steps of S1, target detection; s2, performing recurrent detection; s3, analyzing real-time data; and S4, giving an alarm. According to the method, the advanced computer vision technology is adopted through equipment, efficient and accurate tracking type detection is focused on large facilities, the working process is designed to be quite scientific and careful, firstly, specific targets are arranged, the targets serve as basic points of detection, the accuracy and reliability of the detection process can be ensured, and then the accuracy and reliability of the detection process can be ensured; the device can perform multi-round target recognition, the process not only enhances the reliability of data, but also improves the detection precision, the device can further form an overall displacement cloud picture on the basis of multi-round target recognition, and the cloud picture visually shows the displacement change conditions of large facilities such as side slopes in different time periods. And the problem of identification fault tolerance of a single target spot can be effectively solved.

Description

Technical Field

[0001] The invention relates to the field of monitoring technology, and in particular to a landslide early warning method based on multi-modal target tracking and intelligent recognition. Background Art

[0002] In the field of transportation, especially in the monitoring of highway slopes, bridges, buildings, etc., accurate target identification and monitoring are crucial to ensuring safety. With the development of technology, vision-based monitoring methods have gradually been applied, but there are still many problems. First, the monitoring range is limited. Traditional detection equipment mostly uses fixed bases. When facing long highway slopes, bridges and other large structures, it is difficult to meet the comprehensive and efficient detection needs. Secondly, the monitoring accuracy will be affected by environmental factors and the accuracy limitations of the equipment itself. Some monitoring technologies are greatly affected by environmental factors, resulting in monitoring data errors and affecting accuracy. In addition, some monitoring equipment itself has limited accuracy and insufficient monitoring capabilities for small deformations. Small deformations of highway slopes may be early signs of potential disasters, but existing equipment is difficult to capture accurately. There are also data processing and transmission problems in the current application of monitoring technology.

[0003] The existing data processing efficiency is low. Deformation monitoring will generate a large amount of data. The existing data processing algorithms and systems are inefficient when processing massive amounts of data, making it difficult to analyze and process in real time or quickly, and unable to detect deformation problems in a timely manner. The data transmission is also unstable. The data transmission link in the monitoring system is susceptible to external interference, resulting in unstable or interrupted data transmission, affecting the real-time and integrity of the monitoring data. In remote mountainous areas or complex terrain environments, transmission problems are more prominent. Finally, the existing monitoring technology system is not reliable enough. On the one hand, dynamic monitoring is unreliable, and some equipment lacks reliability in dynamic measurement. It is impossible to guarantee accurate and stable acquisition of deformation data under dynamic conditions such as complex traffic loads and environmental changes. On the other hand, it is manifested in weak anti-interference ability and poor anti-environmental interference performance, which can easily lead to inaccurate or unstable monitoring data due to external environmental factors. Summary of the invention

[0004] In view of the shortcomings of the existing technology, the present invention provides a landslide early warning method based on multi-modal target tracking and intelligent identification, which solves the problems of inaccurate target identification, limited monitoring range, insufficient precision, and unstable data processing and transmission in the existing monitoring technology, and can effectively solve the problem of fault tolerance in the identification of a single target.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a landslide early warning method based on multi-modal target tracking and intelligent identification, specifically comprising the following steps:

[0006] S1. Target detection: The detection equipment deeply integrates the deformation monitoring system to perform high-precision target detection;

[0007] S2. Cyclic detection: The movable base performs continuous and periodic detection on the targets within the visible range. During the cyclic detection process, the movable base moves according to the preset trajectory and speed, and scans and identifies each target one by one. Each scan will form an independent data point, which is transmitted in real time through the signal transmission device for subsequent data processing and analysis.

[0008] S3. Real-time data analysis: Use multi-target trackers to detect, identify, transmit and process targets. Use real-time transmission technology to quickly transmit the data set obtained from each round of detection to the analysis system. After preliminary processing, the data set generates a dynamic cloud map.

[0009] S4. Issue a warning: Through analysis, the system will mark the base section where collapse may occur and immediately trigger an alarm at the main controller.

[0010] Preferably, in S1, the detection device realizes accurate detection of the target and quickly transmits relevant data, which not only includes the position information of the target, but also covers key parameters such as its deformation degree.

[0011] Preferably, in S2, since each cycle will form a complete data group, the data of multiple cycles are compared and analyzed, and more complex algorithms and techniques are used to ensure that the targets in different frames can be accurately associated, involving multiple steps such as extraction, matching and tracking of target features. By combining these steps with cycle detection, a complete and dynamic cloud map is formed, which intuitively displays the motion trajectories and relationships of multiple targets at different time points and spatial positions.

[0012] Preferably, in S3, the cloud map can intuitively display the position and deformation of the target at different time points, providing an important visual reference for subsequent real-time analysis and comparison.

[0013] Preferably, in said S3, during the real-time analysis and comparison process, the data in the cloud map is deeply mined and processed to form an overall deformation map within the base segment. This deformation map not only shows the overall deformation trend of the target over a period of time, but also reveals possible abnormal deformation points, providing timely warning for the maintenance and maintenance of the facility.

[0014] Preferably, in S1, the detection device is a target device, and the target has multiple target recognition objects with multiple colors and shapes that can adapt to different light waves, and the targets of different shapes have unique geometric features.

[0015] Preferably, in S2, the base of the movable platform adopts advanced mobile technology and structural design, so that the detection equipment can easily cope with various complex terrains and detection environments, can quickly and accurately move to the designated detection position, and perform comprehensive and detailed scanning and detection of structures such as slopes.

[0016] Preferably, in S3, a plurality of standard target positions with known positions are set in the detection area. These target positions can be used as reference points to accurately determine the position and posture of the device during movement. During the detection process, the device will scan these standard target positions in turn, and compare the collected data with the preset target position information to calculate the current position and posture deviation of the device. Then, these data are used to correct the collected point cloud image to offset the displacement error caused by movement.

[0017] Preferably, a multi-point automatic tracking algorithm is introduced in S2, and an improved SIFT and LBP combined feature is used to perform non-overlapping irregular block processing on the acquired image, and the SLIC segmentation algorithm is processed by using an adaptive initial value method; SIFT features are extracted from the obtained image blocks, and rotationally invariant uniform LBP features are extracted in the SIFT feature area to be described to establish a feature description.

[0018] Beneficial Effects

[0019] The present invention provides a landslide warning method based on multi-modal target tracking and intelligent identification. Compared with the prior art, it has the following beneficial effects: the landslide warning method based on multi-modal target tracking and intelligent identification uses advanced computer vision technology through the equipment, and focuses on efficient and accurate tracking detection of large facilities. Its work flow is designed to be quite scientific and careful. First, by setting specific targets, these targets serve as the basic points of detection, which can ensure the accuracy and reliability of the detection process. Subsequently, the equipment will perform multiple cycles of target recognition. This process not only enhances the reliability of the data, but also improves the accuracy of detection. On the basis of multiple cycles of target recognition, the equipment can further form an overall displacement cloud map. This cloud map intuitively shows the displacement changes of large facilities such as slopes in different time periods, and provides important data support for subsequent deformation detection. In order to more deeply analyze the deformation of the facilities, the equipment will also generate The displacement cloud map is compared with the preset database, so as to timely discover abnormal deformation of the facility and provide timely warning for the maintenance and upkeep of the facility. It is worth mentioning that the equipment is based on the original detection instrument and cleverly sets up a movable platform. The design of this platform is quite flexible, including but not limited to rotating wheel sets, mobile wheel sets, and hovering drone gimbals. Such a design enables the equipment to easily cope with various complex terrains and detection environments, fundamentally solving the problem that traditional detection methods are difficult to control over a large range. The equipment not only has high-precision detection capabilities, but also has excellent flexibility and adaptability. It can conduct all-round and multi-angle tracking detection of large facilities, timely discover the deformation of facilities, provide strong guarantees for the safe operation of facilities, and effectively solve the recognition fault tolerance problem of single targets. It is believed that this equipment will play an increasingly important role in future facility detection and maintenance work. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is the principle diagram of the detection design of the present invention;

[0021] Figure 2 It is the principle diagram of the detection device identification of the present invention;

[0022] Figure 3 This is a schematic diagram of the target structure of the present invention;

[0023] Figure 4 This is a flow chart of the automatic tracking algorithm of the present invention. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of 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.

[0025] See also Figure 1-3 The present invention provides a technical solution: a landslide early warning method based on multi-modal target tracking and intelligent identification, which specifically includes the following steps:

[0026] S1. Target detection: The detection equipment deeply integrates the deformation monitoring system to perform high-precision target detection;

[0027] S2. Cyclic detection: The movable base performs continuous and periodic detection on the targets within the visible range. During the cyclic detection process, the movable base moves according to the preset trajectory and speed, and scans and identifies each target one by one. Each scan will form an independent data point, which is transmitted in real time through the signal transmission device for subsequent data processing and analysis.

[0028] S3. Real-time data analysis: Use multi-target trackers to detect, identify, transmit and process targets. Use real-time transmission technology to quickly transmit the data set obtained from each round of detection to the analysis system. After preliminary processing, the data set generates a dynamic cloud map.

[0029] S4. Issue a warning: Through analysis, the system will mark the base section where collapse may occur and immediately trigger an alarm at the main controller.

[0030] Specifically, target detection is the cornerstone of visual multi-target tracking technology. Its core task is to accurately locate and identify targets from images or video frames. This technology is crucial for achieving efficient and accurate visual monitoring and analysis, and can capture and analyze tiny deformations of target objects in real time.

[0031] Specifically, cyclic detection is an efficient and reliable detection method that uses a movable base to perform continuous and periodic detection of targets within the visible range. This detection method not only improves the detection accuracy, but also ensures the continuity and integrity of the data.

[0032] Specifically, real-time data analysis is the core requirement of multi-target trackers in complex video stream processing. Since the video stream is continuous, the multi-target tracker must complete target detection, identification, data transmission and processing in a very short time to ensure the real-time and accuracy of the analysis.

[0033] In the present invention, in S1, the detection device realizes accurate detection of the target and quickly transmits the relevant data, which not only includes the position information of the target, but also covers key parameters such as its deformation degree.

[0034] In the present invention, in S2, since each cycle will form a complete data group, the data of multiple cycles are compared and analyzed, and more complex algorithms and techniques are used to ensure that the targets in different frames can be accurately associated, involving multiple steps such as extraction, matching and tracking of target features. By combining these steps with cycle detection, a complete and dynamic cloud map is formed, which intuitively displays the motion trajectories and mutual relationships of multiple targets at different time points and spatial positions.

[0035] In the present invention, in S3, the cloud map can intuitively display the position and deformation of the target at different time points, providing an important visual reference for subsequent real-time analysis and comparison.

[0036] In the present invention, in S3, in the process of real-time analysis and comparison, the data in the cloud map is deeply mined and processed to form an overall deformation map within the base segment. This deformation map not only shows the overall deformation trend of the target over a period of time, but also reveals possible abnormal deformation points, providing timely warning for the maintenance and maintenance of the facility.

[0037] Specifically, real-time data analysis is the key capability of multi-target trackers in complex video stream processing. It ensures the real-time and accuracy of data, and provides strong support for subsequent deformation analysis and early warning.

[0038] In the present invention, in S1, the detection device is a target device, and the target has multiple target recognition objects with multiple colors and shapes that can adapt to different light waves, and the targets of different shapes have unique geometric features.

[0039] In the present invention, in S2, the base of the movable base adopts advanced mobile technology and structural design, so that the detection equipment can easily cope with various complex terrains and detection environments, can quickly and accurately move to the designated detection position, and perform comprehensive and detailed scanning and detection of structures such as slopes.

[0040] In the present invention, in S3, a plurality of standard target positions with known positions are set in the detection area. These target positions can be used as reference points to accurately determine the position and posture of the device during movement. During the detection process, the device will scan these standard target positions in turn, and compare the collected data with the preset target position information to calculate the current position and posture deviation of the device. Then, these data are used to correct the collected point cloud image to offset the displacement error caused by movement.

[0041] Specifically, targets of different shapes (such as crosses, polygons, etc.) have unique geometric features, which can provide rich information in the target recognition process. For example, cross-shaped targets have obvious horizontal and vertical cross structures, and polygonal targets have multiple edges and angles. These unique shape features can help the recognition algorithm identify targets more accurately and reduce misjudgments. In complex monitoring environments, targets of different shapes can avoid recognition confusion caused by similar shapes. For example, in a scene with multiple similar targets, by using targets of different shapes, each target can have a unique identification, which is convenient for accurate distinction. In addition, targets of different shapes present different visual features at different angles. For example, a circular target looks circular at any angle, while a polygonal target will present different combinations of edges and angles at different angles. This makes it possible to improve the overall recognition accuracy by combining targets of multiple shapes during multi-angle monitoring; it can effectively solve the recognition fault tolerance problem of a single target.

[0042] Targets of different appearances, shapes and colors can be selected according to different usage scenarios and monitoring objectives. In some monitoring scenarios, targets of specific shapes may be more suitable. For example, in an environment with a large number of circular interferences (such as pipelines, wellheads, etc.), square or polygonal targets can be more easily distinguished from the surrounding environment, while in an environment with more square structures, circular or irregularly shaped targets may be more recognizable.

[0043] In the present invention, a multi-point automatic tracking algorithm is introduced in S2, and an improved SIFT and LBP combined feature is used to perform non-overlapping irregular block processing on the acquired image, and the SLIC segmentation algorithm is processed by using an adaptive initial value method; SIFT features are extracted from the obtained image blocks, and rotationally invariant uniform LBP features are extracted in the SIFT feature area to be described to establish a feature description.

[0044] Specifically, in S4, the alarm can quickly attract the attention of relevant personnel so that they can take flow limiting measures in time and maintain the area, thereby effectively avoiding the occurrence of landslide accidents and ensuring safety.

[0045] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0046] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0047] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A landslide early warning method based on multi-modal target tracking and intelligent identification, characterized in that: The specific steps include: S1. Target detection: The detection equipment deeply integrates the deformation monitoring system to perform high-precision target detection; S2. Cyclic detection: The movable base performs continuous and periodic detection on the targets within the visible range. During the cyclic detection process, the movable base moves according to the preset trajectory and speed, and scans and identifies each target one by one. Each scan will form an independent data point, which is transmitted in real time through the signal transmission device for subsequent data processing and analysis. S3. Real-time data analysis: Use multi-target trackers to detect, identify, transmit and process targets. Use real-time transmission technology to quickly transmit the data set obtained from each round of detection to the analysis system. After preliminary processing, the data set generates a dynamic cloud map. S4. Issue a warning: Through analysis, the system will mark the base section where collapse may occur and immediately trigger an alarm at the main controller.

2. The landslide early warning method based on multi-modal target tracking and intelligent identification according to claim 1 is characterized by: In S1, the detection equipment realizes accurate detection of the target and quickly transmits the relevant data, which not only includes the position information of the target, but also covers key parameters such as its deformation degree.

3. The landslide early warning method based on multi-modal target tracking and intelligent identification according to claim 1 is characterized by: In S2, since each cycle will form a complete data group, the data of multiple cycles are compared and analyzed, and more complex algorithms and technologies are used to ensure that the targets in different frames can be accurately associated, involving multiple steps such as extraction, matching and tracking of target features. By combining these steps with cycle detection, a complete and dynamic cloud map is formed, which intuitively displays the motion trajectories and relationships of multiple targets at different time points and spatial positions.

4. The landslide early warning method based on multi-modal target tracking and intelligent identification according to claim 1 is characterized by: In S3, the cloud map can intuitively display the position and deformation of the target at different time points, providing an important visual reference for subsequent real-time analysis and comparison.

5. The landslide early warning method based on multi-modal target tracking and intelligent identification according to claim 1 is characterized by: In the S3, during the real-time analysis and comparison process, the data in the cloud map is deeply mined and processed to form an overall deformation map within the base segment. This deformation map not only shows the overall deformation trend of the target over a period of time, but also reveals possible abnormal deformation points, providing timely warnings for the maintenance and upkeep of the facilities.

6. The landslide early warning method based on multi-modal target tracking and intelligent identification according to claim 1 is characterized by: In S1, the detection device is a target device, and the target has multiple target recognition objects with multiple colors and shapes that can adapt to different light waves, and the targets of different shapes have unique geometric features.

7. The landslide early warning method based on multi-modal target tracking and intelligent identification according to claim 1 is characterized by: In the S2, the base of the movable platform adopts advanced mobile technology and structural design, so that the detection equipment can easily cope with various complex terrains and detection environments, can quickly and accurately move to the designated detection position, and perform comprehensive and detailed scanning and detection of structures such as slopes.

8. The landslide early warning method based on multi-modal target tracking and intelligent identification according to claim 1 is characterized by: In S3, a plurality of standard target positions with known positions are set in the detection area. These target positions can be used as reference points to accurately determine the position and posture of the device during movement. During the detection process, the device will scan these standard target positions in turn and compare the collected data with the preset target position information to calculate the current position and posture deviation of the device. Then, these data are used to correct the collected point cloud image to offset the displacement error caused by movement.

9. The landslide early warning method based on multi-modal target tracking and intelligent identification according to claim 1 is characterized by: The S2 introduces a multi-point automatic tracking algorithm, adopts an improved SIFT and LBP combined feature, divides the acquired image into non-overlapping irregular blocks, and uses an adaptive initial value method to divide the SLIC segmentation algorithm into blocks; extracts SIFT features from the obtained image blocks, extracts rotationally invariant uniform LBP features in the SIFT feature area to be described, and establishes feature description.

Images