A three-dimensional laser scanner measurement method and system

Abstract

The present invention proposes a three-dimensional laser scanner measurement method and system. Belonging to the field of measurement technology, the method includes: initializing the device, the initialization setting includes arranging multiple LED marking points on the measured object, starting the LED optical tracking system, and setting the parameters of the three-dimensional laser scanner; acquiring a video stream containing LED marking points in real time according to the LED optical tracking system, and transmitting the video stream to the cloud space for real-time processing, identifying and extracting the position information of each LED marking point, and calculating its motion trajectory and speed according to the position change of the marking point. The measurement accuracy of three-dimensional laser scanning can be significantly improved by accurately tracking the LED marking points and dynamically adjusting the scanning path. This method can correct the scanning path in real time, so that the laser beam is always focused on the key area, reducing the error caused by the movement of the object.

Description

Technical Field

[0001] The invention provides a three-dimensional laser scanner measurement method and system, belonging to the technical field of measurement. Background Art

[0002] In the fields of modern industrial manufacturing, cultural relics protection, medical diagnosis, and entertainment film and television production, the precise construction and dynamic tracking of three-dimensional models have become one of the key technologies. Although traditional three-dimensional measurement methods, such as contact measurement, structured light scanning, or single laser scanning, can meet the needs of three-dimensional reconstruction of static objects to a certain extent, they often have problems such as low scanning efficiency, incomplete data acquisition, and susceptibility to environmental interference when facing complex and dynamically changing objects. Especially in scenarios where real-time monitoring and reconstruction of moving objects are required, the limitations of these methods are more significant.

[0003] In order to overcome these challenges, optical tracking technology has emerged, especially optical tracking technology based on LED (light-emitting diode), which has gradually become a research hotspot in the field of dynamic 3D measurement due to its advantages of high brightness, low energy consumption, and fast response speed. LED markers are widely used in machine vision, motion capture, augmented reality and other fields due to their easy arrangement, high recognition rate, and stable operation under various lighting conditions.

[0004] However, traditional LED optical tracking systems are mostly limited to real-time capture of position information, and have not effectively combined with high-precision 3D scanning technology to achieve complete acquisition of surface details of objects. At the same time, 3D scanning in dynamic environments often faces problems such as unreasonable scanning path planning, large amount of data redundancy, and low real-time processing efficiency, making it difficult to meet the high requirements for real-time, accuracy, and robustness in practical applications.

[0005] Therefore, a 3D laser scanner measurement method is studied, which aims to dynamically adjust the scanning path of the laser scanner through real-time tracking and position prediction of LED markers, so as to achieve efficient and accurate 3D reconstruction of the surface of dynamic objects. This method can not only significantly improve scanning efficiency and reduce data redundancy, but also effectively cope with the dynamic changes of the measured object, providing more accurate and efficient 3D modeling solutions for industrial inspection, animation production, virtual reality and other fields. Summary of the invention

[0006] The present invention provides a three-dimensional laser scanner measurement method and system to solve the problems mentioned in the above background technology:

[0007] The present invention provides a three-dimensional laser scanner measurement method, the method comprising:

[0008] S1, initializing the device, wherein the initialization setting includes arranging multiple LED marking points on the object to be measured, starting the LED optical tracking system, and setting the parameters of the three-dimensional laser scanner;

[0009] S2, according to the LED optical tracking system, real-time acquisition of the video stream containing the LED marking points, and transmission of the video stream to the cloud space for real-time processing, identification and extraction of the position information of each LED marking point, and calculation of its motion trajectory and speed according to the position change of the marking point;

[0010] S3, dynamically adjust the scanning path of the 3D laser scanner based on the position and motion trajectory of the LED marker point; and update the scanning path in real time according to the marker point position information obtained in real time;

[0011] S4, start the 3D laser scanner, scan according to the planned scanning path, and collect the 3D coordinate data of all scanning points to form a 3D point cloud of the surface of the object to be measured;

[0012] S5. Align the marker point position information obtained by the LED optical tracking system with the point cloud data obtained by the 3D laser scanner; establish a mapping relationship between the two by calculating the corresponding position of the marker point in the 3D point cloud; combine the 3D point cloud data at different time points with the timestamp information to form a dynamic 3D model of the object being measured.

[0013] Furthermore, the S1 includes:

[0014] S11. According to the shape, size and expected range of motion of the object to be measured, the distribution of the marking points is simulated by computer simulation software, and the LED marking points are fixed at the key positions of the object to be measured by a high-precision mechanical arm;

[0015] S12, uniquely number each LED marking point and record its initial position information through a static calibration procedure;

[0016] S13, after starting the system, perform self-test and preheating at the same time; adjust the parameters of the camera according to the size, moving speed and ambient light conditions of the object to be measured;

[0017] S14. Select the scanning accuracy and scanning speed according to the measurement requirements, and balance the measurement accuracy and measurement time; set the scanning range and scanning angle of the laser scanner according to the size and shape of the object to be measured; and configure the data transmission protocol and storage path.

[0018] Furthermore, the S2 includes:

[0019] S21, continuously capture the video stream containing the LED marker points through a high frame rate camera; perform real-time compression processing on the video stream and transmit it to the cloud space through wired or wireless means;

[0020] S22, the cloud space pre-processes the received video frames and identifies the LED marking points in the video through a machine learning algorithm;

[0021] S23, based on the recognition result, extract the pixel coordinates of each marking point, and convert them into three-dimensional space coordinates through camera calibration parameters;

[0022] S24. According to the position information of the marker point in the continuous frames, the Kalman filter algorithm is used to track its motion trajectory; based on the trajectory data, the instantaneous velocity and acceleration of the marker point are calculated.

[0023] Furthermore, the S21 includes:

[0024] The video stream containing the LED markers is captured by a high-frequency camera. During the video capture process, the video stream is previewed in real time, and the camera parameters are dynamically adjusted according to the preview effect.

[0025] Through the video encoding algorithm, the captured video stream is efficiently compressed. According to the dynamic changes of the video content, based on the preset intelligent compression strategy, the key frames and key areas are encoded with high quality, and the non-key frames and background areas are compressed with low quality.

[0026] During the compression process, the video frames are preprocessed by real-time denoising and contrast enhancement algorithms;

[0027] According to the on-site environment and network conditions, select a combination of wired and wireless transmission methods to transmit the video stream;

[0028] Using multi-path transmission technology, the video stream is divided into multiple data packets and transmitted in parallel to the cloud space through different network paths.

[0029] Furthermore, the S23 includes:

[0030] Based on the recognition of LED markers by machine learning algorithms, the edges of the markers are further refined through morphological operations;

[0031] Through sub-pixel positioning technology, pixel coordinates are extracted on the refined edges;

[0032] Through the camera calibration method, combined with the standard calibration plate or three-dimensional calibration target, the camera's intrinsic parameter matrix and extrinsic parameter matrix are obtained;

[0033] According to the calibration results, the image in the video frame is distorted and corrected, and the perspective projection transformation matrix from the pixel coordinate system to the camera coordinate system is constructed based on the pinhole camera model;

[0034] Combined with the position and posture of the camera in the world coordinate system, the coordinates in the camera coordinate system are converted to the world coordinate system;

[0035] If the LED marker itself does not contain depth information (such as ordinary LED lights), multi-viewing, structured light or ToF sensors are used to assist in depth estimation;

[0036] The depth estimation results are combined with the perspective projection model to optimize the calculation of three-dimensional space coordinates, and the various error sources that affect the accuracy of coordinate transformation are analyzed. Based on the identified error sources, the corresponding correction method is used to correct the coordinate transformation results.

[0037] Furthermore, the S3 includes:

[0038] S31, using a path optimization algorithm based on a prediction model, predicting the future position according to the movement trend of the marker point, planning the optimal scanning path; and smoothing the planned path;

[0039] S32, based on the real-time data feedback mechanism, feeding back the latest position information of the LED tracking system to the scanning path planning system in real time;

[0040] S33. According to the real-time position information, the scanning path is dynamically adjusted so that the laser beam is always focused on the key part of the measured area.

[0041] Furthermore, the S31 includes:

[0042] Obtain historical data, analyze and process the acquired historical data, select features based on the motion characteristics of the LED markers, extract the selected features, and construct the feature space of the prediction model;

[0043] Through deep learning algorithms, historical data and feature space are combined to train a model that can predict the future position of LED markers;

[0044] Use the trained prediction model to predict the future position of the LED marker and perform uncertainty assessment on the prediction results;

[0045] According to the specific needs of the application scenario, the objective function of the scanning path is set, and the optimal scanning path is planned through the path optimization algorithm combined with the predicted future position and uncertainty assessment results;

[0046] Perform curvature analysis on the planned path to identify possible sharp change points or inflection points in the path; use a smoothing algorithm to smooth the sharp change points in the path.

[0047] Furthermore, the S4 includes:

[0048] S41, according to the adjusted scanning path, start the 3D laser scanner to perform the scanning task, monitor the scanning process, and receive the returned scanning data;

[0049] S42, collecting three-dimensional coordinate data of all scanning points, wherein the three-dimensional coordinate data includes X, Y, Z coordinates and reflection intensity information;

[0050] S43, performing quality check on the collected raw data, removing noise points and outliers; filtering and smoothing the valid data;

[0051] S44, and organize the processed three-dimensional coordinate data into a point cloud format.

[0052] Furthermore, the S5 includes:

[0053] S51, extracting common features from the marker point positions obtained by the LED optical tracking system and the point cloud obtained by the three-dimensional laser scanner;

[0054] S52, using iterative nearest point, matching the position of the marker point with the corresponding feature in the point cloud; and calculating the transformation matrix from the marker point coordinate system to the point cloud coordinate system according to the matching result;

[0055] S53, aligning the three-dimensional point cloud data from different time points according to the timestamps, and fusing the point cloud data of adjacent time points through a sliding window according to the continuity and smoothness of the object's motion, to generate a continuous dynamic point cloud sequence;

[0056] S54, using the point cloud data to construct a surface geometric model of the object being measured through triangulation; dynamically adjusting and updating the three-dimensional model according to the point cloud data updated in real time;

[0057] S55. Optimize the constructed dynamic three-dimensional model and display the dynamic three-dimensional model through three-dimensional visualization software or platform.

[0058] The present invention provides a three-dimensional laser scanner measurement system, the system comprising:

[0059] Initialization setting module: initializing the device, including arranging multiple LED marking points on the object to be measured, starting the LED optical tracking system, and setting the parameters of the 3D laser scanner;

[0060] Data processing module: The LED optical tracking system obtains the video stream containing the LED marker points in real time, transmits the video stream to the cloud space for real-time processing, identifies and extracts the position information of each LED marker point, and calculates its motion trajectory and speed according to the position change of the marker point;

[0061] Path update module: dynamically adjust the scanning path of the 3D laser scanner based on the position and motion trajectory of the LED marker points; and update the scanning path in real time based on the marker point position information obtained in real time;

[0062] Point cloud generation module: Start the 3D laser scanner, scan according to the planned scanning path, and collect the 3D coordinate data of all scanning points to form a 3D point cloud of the surface of the object being measured;

[0063] Model generation module: align the marker point position information obtained by the LED optical tracking system with the point cloud data obtained by the 3D laser scanner; establish a mapping relationship between the two by calculating the corresponding positions of the marker points in the 3D point cloud; combine the 3D point cloud data at different time points with the timestamp information to form a dynamic 3D model of the object being measured.

[0064] Beneficial effects of the present invention: The measurement accuracy of three-dimensional laser scanning can be significantly improved by accurately tracking the LED marker points and dynamically adjusting the scanning path. This method can correct the scanning path in real time, so that the laser beam is always focused on the key area, reducing the error caused by the movement of the object; combining the data of the LED optical tracking system and the three-dimensional laser scanner, a dynamic three-dimensional model of the measured object can be generated. This is particularly important for application scenarios that require dynamic monitoring or analysis (such as industrial inspection, motion analysis, etc.); using cloud computing and high frame rate cameras, real-time processing and transmission of video streams ensures the instant acquisition and update of the position of the LED marker points. This enables instant tracking and rapid response to the movement of objects, improving the real-time performance of the system; by planning and optimizing the scanning path through predictive models and deep learning algorithms, blind spots and repeated scanning during the scanning process can be effectively avoided, improving scanning efficiency and coverage of point cloud data; in the data collection and processing stage, the quality of the final point cloud data is ensured through noise filtering, outlier removal, data smoothing and other steps, avoiding noise introduced by environmental interference or measurement errors; by aligning the position data obtained by the LED optical tracking system with the point cloud data of the laser scanner, the point cloud data at different time points can be accurately integrated into a complete dynamic model, enhancing the continuity and accuracy of the model; the final constructed dynamic 3D model not only has high precision and resolution, but can also be displayed and analyzed through 3D visualization software, providing strong support for subsequent engineering applications, design optimization or scientific research. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a step diagram of the method of the present invention;

[0066] Figure 2 This is a system module diagram of the present invention. DETAILED DESCRIPTION

[0067] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0068] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. The embodiments described are only a part of the embodiments of the present invention, rather than 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.

[0069] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0070] One embodiment of the present invention, as Figure 1 As shown, a three-dimensional laser scanner measurement method, the method comprising:

[0071] S1, initializing the device, wherein the initialization setting includes arranging multiple LED marking points on the object to be measured, starting the LED optical tracking system, and setting the parameters of the three-dimensional laser scanner;

[0072] S2, according to the LED optical tracking system, real-time acquisition of the video stream containing the LED marking points, and transmission of the video stream to the cloud space for real-time processing, identification and extraction of the position information of each LED marking point, and calculation of its motion trajectory and speed according to the position change of the marking point;

[0073] S3. Dynamically adjust the scanning path of the 3D laser scanner based on the position and motion trajectory of the LED marker point; specifically, plan the scanning path as a path around or following the marker point to ensure that the laser beam can fully cover the measured area and reduce invalid scanning; and update the scanning path in real time according to the marker point position information obtained in real time to cope with the dynamic changes of the measured object;

[0074] S4, start the 3D laser scanner, scan according to the planned scanning path, and collect the 3D coordinate data of all scanning points to form a 3D point cloud of the surface of the object to be measured;

[0075] S5. Align the marker point position information obtained by the LED optical tracking system with the point cloud data obtained by the 3D laser scanner; establish a mapping relationship between the two by calculating the corresponding position of the marker point in the 3D point cloud; combine the 3D point cloud data at different time points with the timestamp information to form a dynamic 3D model of the object being measured.

[0076] The working principle of the above technical solution is as follows: multiple LED markers are arranged on the object to be measured, and these markers serve as reference points of the tracking system to reflect the position and posture of the object in real time; the tracking system is started, and the system captures the video stream containing the LED markers through sensors such as cameras; according to the measurement requirements, the relevant parameters of the three-dimensional laser scanner, such as scanning speed, resolution, scanning range, etc., are set; the LED optical tracking system obtains the video stream containing the LED markers in real time, and transmits these video streams to the cloud space for centralized processing. The cloud space provides powerful computing power and can process a large amount of data in real time; in the cloud space, the position information of each LED marker is identified and extracted through image processing algorithms. These position information include the pixel coordinates of the marker in the image, and the three-dimensional coordinates in the world coordinate system calculated by camera calibration and perspective transformation; according to the changes in the position of the marker in consecutive frames, its motion trajectory and speed are calculated. This helps to predict the future position of the marker and provide a basis for subsequent scanning path planning; based on the position and motion trajectory of the LED marker, the scanning path of the three-dimensional laser scanner is dynamically planned. The path planning algorithm plans the scanning path as a path around or following the marker point to ensure that the laser beam can fully cover the measured area, while reducing invalid scanning and improving scanning efficiency. The scanning path is updated in real time according to the real-time acquired marker point position information. This can cope with the dynamic changes of the measured object and ensure the accuracy and real-time performance of the scanning; start the 3D laser scanner for scanning according to the planned scanning path; during the scanning process, the 3D laser scanner will collect the 3D coordinate data of all scanning points. These data reflect the geometric shape and position information of the surface of the measured object; the collected 3D coordinate data is sorted into a point cloud format to form a 3D point cloud of the surface of the measured object. The point cloud is a collection of a series of discrete points in a 3D space that can intuitively display the 3D form of the object; the marker point position information obtained by the LED optical tracking system is registered with the point cloud data obtained by the 3D laser scanner. By calculating the corresponding position of the marker point in the 3D point cloud, a mapping relationship between the two is established. This helps to combine the position information of the tracking system with the scanning data to improve the accuracy of the 3D model; combined with the timestamp information, the 3D point cloud data at different time points are integrated. Through time series analysis, a dynamic 3D model of the measured object can be constructed. The model can reflect the morphological changes of objects at different time points and provide a basis for subsequent analysis, simulation and visualization.

[0077] The effects of the above technical solution are as follows: the LED optical tracking system can capture the position information of the LED marker point in real time and accurately, and ensure the accuracy of the position information through high-precision image processing and algorithm optimization; the video stream is transmitted to the cloud space for real-time processing, which can quickly identify and extract the position information of the LED marker point, and calculate the motion trajectory and speed accordingly, providing real-time and reliable data support for subsequent scanning path planning; according to the position and motion trajectory of the LED marker point, the scanning path of the three-dimensional laser scanner is dynamically adjusted to ensure that the laser beam can fully cover the measured area and reduce invalid scanning. This not only improves the scanning efficiency, but also saves scanning time and resources; the real-time acquisition and update of the position information of the marker point can flexibly respond to the dynamic changes of the measured object and ensure the accuracy and real-time performance of the scanning path; the three-dimensional laser scanner can collect the three-dimensional coordinate data of all scanning points to form a high-precision point cloud on the surface of the measured object. These point cloud data can truly reflect the geometric shape and position information of the object; by calculating the corresponding position of the LED marker point in the three-dimensional point cloud, the mapping relationship between the two is established, and the precise registration of the tracking system position information and the scanning data is achieved. By combining the timestamp information and integrating the 3D point cloud data at different time points, a dynamic 3D model of the object being measured can be constructed. This model can reflect the morphological changes of the object at different time points, providing a basis for subsequent analysis, simulation and visualization.

[0078] In one embodiment of the present invention, the S1 includes:

[0079] S11. According to the shape, size and expected range of motion of the object to be measured, the optimal distribution of marking points is simulated by computer simulation software; the LED marking points are fixed at key positions of the object to be measured, such as corners, edges or dynamically changing areas, by high-precision robotic arms to ensure that the marking points are stable and do not affect the function of the object.

[0080] S12, uniquely number each LED marking point and record its initial position information through a static calibration procedure;

[0081] S13, after starting the system, perform self-test, including camera clarity, LED recognition algorithm efficiency, etc., and perform preheating treatment; adjust the camera parameters, including frame rate, exposure time, and image resolution, according to the size, moving speed, and ambient light conditions of the object being measured;

[0082] S14. Select the scanning accuracy and scanning speed according to the measurement requirements, and balance the measurement accuracy and measurement time; set the scanning range and scanning angle of the laser scanner according to the size and shape of the object to be measured; and configure the data transmission protocol and storage path.

[0083] The working principle of the above technical solution is as follows: using computer simulation software, according to the shape, size and expected range of motion of the object to be measured, simulate the best distribution scheme of LED marker points. This step is to ensure that no matter how the object moves, the LED optical tracking system can effectively capture enough marker point information; the LED marker points are accurately fixed at key positions of the object to be measured, such as corners, edges or dynamically changing areas, through high-precision robotic arms. The selection of these positions is based on simulation results and aims to maximize the coverage and accuracy of the tracking system. At the same time, ensure the stability of the marker points to avoid falling off or shifting during the measurement process without affecting the normal function of the object to be measured; assign a unique number to each LED marker point for subsequent data processing and identification; record the initial position information of each marker point through a static calibration procedure. This step provides a reference point for subsequent dynamic tracking, so that the system can accurately calculate the position change of the marker point during the movement; after starting the system, a series of self-test operations are performed, including checking the clarity of the camera, the efficiency of the LED recognition algorithm, etc. At the same time, preheating treatment is carried out to ensure that all components of the system are in the best working state; according to the size, movement speed and ambient light conditions of the object to be measured, the parameters of the camera, such as frame rate, exposure time and image resolution, are adjusted. The optimization of these parameters aims to improve the tracking accuracy and response speed of the system; the scanning accuracy and scanning speed are selected according to the measurement requirements. Under the premise of ensuring the measurement accuracy, the scanning speed is increased as much as possible to shorten the measurement time. At the same time, the measurement accuracy and measurement time are balanced to ensure the optimal allocation of resources under the premise of meeting the measurement requirements; the scanning range and scanning angle of the laser scanner are set according to the size and shape of the object to be measured. This step is to ensure that the laser beam can fully cover the measured area, while avoiding redundant scanning and improving scanning efficiency; configure the data transmission protocol and storage path to ensure that the scanned data can be transmitted to the processing center in real time and efficiently, and stored safely. This step is crucial for subsequent data processing and analysis.

[0084] The effects of the above technical solution are as follows: by simulating the optimal distribution of marker points through computer simulation software, it is ensured that no matter how the measured object moves, the LED optical tracking system can effectively capture sufficient marker point information. This greatly improves the accuracy and reliability of the measurement and reduces the measurement error caused by improper distribution of marker points; the LED marker points are fixed at the key positions of the measured object using a high-precision mechanical arm to ensure the stability of the marker points and avoid measurement failures caused by displacement or falling off of the marker points during the measurement process. At the same time, this fixing method also ensures that the marker points will not affect the normal function of the measured object; each LED marker point is uniquely numbered, and its initial position information is recorded through a static calibration program, providing a reliable benchmark for subsequent dynamic tracking. This numbering and calibration method simplifies the data processing process and improves the accuracy and efficiency of data processing; after starting the system, a comprehensive self-test is performed, including camera clarity, LED recognition algorithm efficiency, etc., to ensure that all components of the system are in the best working condition. At the same time, the camera parameters such as frame rate, exposure time and image resolution are adjusted according to the size, movement speed and ambient lighting conditions of the measured object, which further optimizes the system performance and improves the tracking accuracy and response speed; the scanning accuracy and scanning speed are selected according to the measurement requirements and a balance is made between the two. This selection method not only ensures the measurement accuracy, but also improves the measurement efficiency and reduces unnecessary measurement time; the scanning range and scanning angle of the laser scanner are set according to the size and shape of the measured object to ensure full coverage and avoid redundant scanning. This further improves the measurement efficiency and reduces the burden of data storage and processing; the configuration of efficient data transmission protocols and secure storage paths ensures that the scanned data can be transmitted to the processing center in real time and efficiently, and stored securely. This provides strong support for subsequent data processing and analysis.

[0085] In one embodiment of the present invention, the S2 includes:

[0086] S21, continuously capture a video stream containing LED markers through a high frame rate camera; perform real-time compression processing on the video stream to reduce transmission bandwidth requirements, and transmit the video stream to the cloud space through wired or wireless means;

[0087] S22, the cloud space pre-processes the received video frames, including denoising and contrast enhancement, to improve the accuracy of marker point recognition; and identifies the LED marker points in the video through a machine learning algorithm;

[0088] S23, based on the recognition result, extract the pixel coordinates of each marking point, and convert them into three-dimensional space coordinates through camera calibration parameters;

[0089] S24. According to the position information of the marker point in the continuous frames, the Kalman filter algorithm is used to track its motion trajectory; based on the trajectory data, the instantaneous velocity and acceleration of the marker point are calculated.

[0090] The working principle of the above technical solution is as follows: a video stream containing LED markers is continuously captured by a high frame rate camera. The high frame rate camera can ensure that clear images can be captured even when the object moves quickly, reducing motion blur; the captured video stream is compressed in real time to reduce the transmission bandwidth requirement. The compressed video stream is transmitted to the cloud space for processing by wired or wireless means. This real-time transmission method ensures the timeliness and effectiveness of the data; in the cloud space, the received video frames are preprocessed, including denoising and contrast enhancement. The denoising operation can reduce noise interference in the image and improve image quality; contrast enhancement helps to display the LED markers more clearly and improve recognition accuracy; the LED markers in the preprocessed video frames are identified by machine learning algorithms. The machine learning algorithm can automatically learn and identify the features of the LED markers in the video to achieve fast and accurate recognition; based on the recognition results, the pixel coordinates of each LED marker in the video frame are extracted. These pixel coordinates represent the position of the marker in the two-dimensional image; the pixel coordinates of the marker are converted into three-dimensional space coordinates through camera calibration parameters (such as focal length, optical center position, etc.). This step realizes the mapping from two-dimensional images to three-dimensional space, providing a basis for subsequent three-dimensional scanning and path planning; based on the position information of the LED marker in the continuous frames, the Kalman filter algorithm is used to track its motion trajectory. The Kalman filter algorithm can effectively handle noise and uncertainty, and provide smooth and accurate motion trajectory estimation; based on the tracked motion trajectory data, the instantaneous speed and acceleration of the LED marker are calculated. These parameters reflect the motion state of the marker and provide an important basis for dynamically adjusting the scanning path. For example, when the object moves quickly, the scanning path can be adjusted according to its speed and acceleration to ensure that the laser beam can continue to cover the measured area.

[0091] The effects of the above technical solution are: the video stream is continuously captured by a high frame rate camera, ensuring that a clear and coherent image can be captured even when the object moves quickly. This greatly improves the real-time performance and response speed of the system; the video stream is compressed in real time, which significantly reduces the transmission bandwidth requirement, so that the video data can be quickly and stably transmitted to the cloud space via wired or wireless methods. This efficient transmission method ensures the timeliness and integrity of the data; the cloud space performs pre-processing operations such as denoising and contrast enhancement on the received video frames, effectively improving the image quality and reducing noise interference, thereby improving the recognition accuracy of the LED marker points; the machine learning algorithm is used to identify the LED marker points in the video, realizing an automated and intelligent recognition process. This recognition method is not only highly accurate, but also can adapt to the recognition requirements of marker points under different shapes, sizes and lighting conditions; the pixel coordinates extracted based on the recognition results are converted into three-dimensional space coordinates through the camera calibration parameters. This step realizes the accurate mapping from two-dimensional images to three-dimensional space, providing accurate spatial position information for subsequent three-dimensional scanning and path planning; the Kalman filter algorithm is used to track the position information of the marker points in continuous frames, and a smooth and accurate motion trajectory is obtained. This step not only helps to understand the motion state of the object, but also provides an important basis for dynamically adjusting the scanning path; the instantaneous speed and acceleration of the marker points calculated based on the trajectory data reflect the motion characteristics of the object. This information provides strong support for the dynamic adjustment of the scanning path, allowing the system to flexibly adjust the scanning strategy according to actual conditions to ensure the comprehensiveness and accuracy of the scanning; the entire technical solution has strong adaptability and can cope with objects of different shapes, sizes and motion characteristics. At the same time, through the processing power of the cloud space, functions such as remote monitoring and data analysis can also be realized, further improving the flexibility and scalability of the system.

[0092] In one embodiment of the present invention, the S21 includes:

[0093] The video stream containing the LED markers is captured by a high-frequency camera. During the video capture process, the video stream is previewed in real time, and the camera parameters are dynamically adjusted according to the preview effect.

[0094] Through video encoding algorithms (such as H.265 / HEVC, AV1, etc.), the captured video stream is efficiently compressed. According to the dynamic changes of the video content (such as the movement of the LED marker), based on the preset intelligent compression strategy, the key frames and key areas are encoded with high quality, and the non-key frames and background areas are compressed with low quality.

[0095] During the compression process, the video frames are preprocessed by real-time denoising and contrast enhancement algorithms;

[0096] According to the on-site environment and network conditions, select a combination of wired (such as optical fiber, Ethernet) and wireless (such as Wi-Fi6, 5G) transmission methods to transmit video streams;

[0097] Using multi-path transmission technology, the video stream is divided into multiple data packets and transmitted in parallel to the cloud space through different network paths.

[0098] The working principle of the above technical solution is as follows: a high-frequency camera is used to capture the video stream containing the LED marker. The high-frequency camera can provide smoother video images, especially when capturing fast-moving objects, which can effectively reduce motion blur; during the video capture process, the video stream is previewed in real time, and the camera parameters (such as exposure, focal length, white balance, etc.) are dynamically adjusted according to the preview effect. This real-time feedback mechanism ensures that the video quality can be kept stable and optimized even in complex or changing environments; advanced video encoding algorithms (such as H.265 / HEVC, AV1, etc.) are used to efficiently compress the captured video stream. These algorithms can significantly reduce the amount of data while ensuring video quality, thereby reducing the transmission bandwidth requirements; according to the dynamic changes of the video content (such as the movement of the LED marker), the video is encoded based on a preset intelligent compression strategy. Specifically, key frames and key areas (such as the LED marker and its surrounding areas) are encoded with high quality to retain more detailed information; while non-key frames and background areas are compressed with low quality to save bandwidth; during the compression process, the video frames are pre-processed by real-time denoising and contrast enhancement algorithms. The denoising algorithm can reduce noise interference in the image and improve image clarity; the contrast enhancement algorithm can enhance the contrast between the LED marker and the background, further improving the recognition accuracy of the LED marker; according to the on-site environment and network conditions, a combination of wired (such as optical fiber, Ethernet) and wireless (such as Wi-Fi6, 5G) transmission methods are selected to transmit the video stream. This flexible transmission method can make full use of existing network resources to ensure the stable transmission of the video stream; multi-path transmission technology is used to divide the video stream into multiple data packets, and transmit them to the cloud space in parallel through different network paths. This technology can improve the reliability and fault tolerance of data transmission. Even if some network paths fail, data can continue to be transmitted through other paths to ensure the continuity and integrity of the video stream.

[0099] The effect of the above technical solution is: by previewing the video stream in real time and dynamically adjusting the camera parameters according to the preview effect, it can ensure that high-quality video images can be quickly adapted and captured even in complex or changing environments. This real-time feedback mechanism greatly improves the flexibility and adaptability of the system; the video stream is efficiently compressed using advanced video encoding algorithms (such as H.265 / HEVC, AV1, etc.), significantly reducing the bandwidth required for data transmission. This is particularly important for scenarios where large amounts of video data need to be transmitted in real time, which can reduce network congestion and transmission delays; based on the dynamic changes in video content (such as the movement of LED markers), key frames and key areas are encoded with high quality, and non-key frames and background areas are compressed with low quality. This strategy further reduces the amount of data and optimizes bandwidth usage efficiency while ensuring that key information is not lost; during the compression process, the video frames are pre-processed through real-time denoising and contrast enhancement algorithms, which effectively reduces image noise and improves the contrast between the LED marker and the background. This not only improves the visual effect of the video, but also significantly improves the recognition accuracy of LED markers, providing a more reliable data basis for subsequent three-dimensional spatial positioning and motion trajectory tracking; according to the on-site environment and network conditions, a combination of wired (such as optical fiber, Ethernet) and wireless (such as Wi-Fi6, 5G) transmission methods can be flexibly selected. This flexibility ensures stable and efficient video transmission in different scenarios; multi-path transmission technology is used to divide the video stream into multiple data packets, and transmit them to the cloud space in parallel through different network paths. This technology improves the reliability and fault tolerance of data transmission. Even if some network paths fail, data can continue to be transmitted through other paths to ensure the continuity and integrity of the video stream.

[0100] In one embodiment of the present invention, the S23 includes:

[0101] Based on the recognition of LED markers by machine learning algorithms, the edges of the markers are further refined through morphological operations (such as corrosion, dilation, and edge detection);

[0102] Through sub-pixel positioning technology (such as centroid method, fitting method, interpolation method), pixel coordinates are extracted on the refined edge;

[0103] Obtain the camera's intrinsic and extrinsic matrix through camera calibration methods (such as Zhang Zhengyou's calibration method, Tsai's two-step calibration method, etc.), combined with standard calibration plates or three-dimensional calibration targets;

[0104] According to the calibration results, the image in the video frame is distorted and corrected to eliminate the image distortion caused by lens distortion and improve the spatial geometric accuracy of the image; based on the pinhole camera model, the perspective projection transformation matrix from the pixel coordinate system to the camera coordinate system is constructed;

[0105] Combined with the position and posture of the camera in the world coordinate system (i.e., the external parameter matrix), the coordinates in the camera coordinate system are converted to the world coordinate system; the mapping of pixel coordinates to three-dimensional space coordinates is realized.

[0106] If the LED marker itself does not contain depth information (such as ordinary LED lights), multi-eye vision, structured light or ToF (Time-of-Flight) sensors are used to assist in depth estimation;

[0107] The depth estimation results are combined with the perspective projection model to optimize the calculation of three-dimensional space coordinates. The various error sources that affect the accuracy of coordinate transformation are analyzed, including camera calibration error, distortion correction error, image noise, quantization error, etc. Based on the identified error sources, the corresponding correction methods (such as system calibration optimization, image processing technology, parameter adjustment, etc.) are used to correct the coordinate transformation results.

[0108] The working principle of the above technical solution is as follows: on the basis of the preliminary identification of the LED marker point by the machine learning algorithm, the edge of the marker point is further refined through morphological operations (such as corrosion, expansion, and edge detection). These operations help to remove burrs and noise on the edge, making the edge smoother and more accurate, thereby improving the accuracy of subsequent pixel coordinate extraction; sub-pixel positioning technology (such as centroid method, fitting method, and interpolation method) is used to extract pixel coordinates on the refined edge. Sub-pixel positioning technology can break through the limitations of a single pixel and achieve higher-precision coordinate positioning. This is particularly important for application scenarios that require high-precision three-dimensional spatial positioning. Through camera calibration methods (such as Zhang Zhengyou calibration method, Tsai two-step calibration method, etc.), combined with standard calibration plates or three-dimensional calibration targets, the camera's intrinsic parameter matrix and extrinsic parameter matrix are obtained. The intrinsic parameter matrix contains internal parameters such as the focal length and optical center of the camera, while the extrinsic parameter matrix describes the position and posture of the camera in the world coordinate system. According to the calibration results, the image in the video frame is distorted. Since lens distortion (such as radial distortion and tangential distortion) will cause image distortion and affect the spatial geometric accuracy of the image, distortion correction is needed to eliminate this effect. After distortion correction, straight lines in the image will remain straight lines and parallel lines will remain parallel. Based on the pinhole camera model, a perspective projection transformation matrix from the pixel coordinate system to the camera coordinate system is constructed. This matrix describes the mapping relationship from the two-dimensional image plane to the three-dimensional camera coordinate system. Combined with the position and posture of the camera in the world coordinate system (i.e., the external parameter matrix), the coordinates in the camera coordinate system are converted to the world coordinate system. In this way, the mapping from the pixel coordinates in the video frame to the three-dimensional space coordinates is realized. If the LED marker point itself does not contain depth information (such as ordinary LED lights), depth estimation is required through auxiliary means such as multi-eye vision, structured light or ToF (Time-of-Flight) sensors. These technologies can measure the distance from the object to the camera, thereby providing the necessary depth information. Combining the depth estimation results with the perspective projection model can further optimize the calculation of the three-dimensional space coordinates. At the same time, various error sources that affect the accuracy of coordinate transformation are analyzed, including camera calibration error, distortion correction error, image noise, quantization error, etc. Based on the identified error sources, corresponding correction methods (such as system calibration optimization, image processing technology, parameter adjustment, etc.) are used to correct the coordinate transformation results to improve the final three-dimensional space coordinate accuracy.

[0109] The effects of the above technical solution are as follows: by refining the edge of the LED marker through morphological operations (erosion, expansion, edge detection), the noise and burrs on the edge can be effectively removed, making the edge clearer and more precise. This directly improves the accuracy of subsequent pixel coordinate extraction and lays a solid foundation for subsequent three-dimensional space coordinate mapping; sub-pixel positioning technology (centroid method, fitting method, interpolation method, etc.) is used to extract pixel coordinates on the refined edge, which can break through the limitation of a single pixel and achieve higher-precision coordinate positioning. This is particularly important for application scenarios that require high-precision three-dimensional space positioning; the camera's intrinsic parameter matrix and extrinsic parameter matrix are obtained through the camera calibration method, and the image in the video frame is distorted according to the calibration results. This eliminates the image distortion caused by lens distortion and improves the spatial geometric accuracy of the image. The corrected image is closer to the real scene, providing a more reliable basis for subsequent three-dimensional space coordinate mapping; the perspective projection transformation matrix is ​​constructed based on the pinhole camera model, and the coordinate conversion from the camera coordinate system to the world coordinate system is realized by combining the position and posture of the camera in the world coordinate system (extrinsic parameter matrix). This process maps the pixel coordinates in the video frame into high-precision three-dimensional space coordinates, meeting the needs of applications such as three-dimensional scanning and dynamic tracking; for LED markers that do not contain depth information (such as ordinary LED lights), depth estimation is performed through auxiliary means such as multi-eye vision, structured light or ToF sensors. This provides the necessary depth information, making the calculation of three-dimensional space coordinates more complete and accurate; the depth estimation results are combined with the perspective projection model to optimize the calculation of three-dimensional space coordinates. At the same time, various error sources that affect the accuracy of coordinate conversion are analyzed, and corresponding correction methods (such as system calibration optimization, image processing technology, parameter adjustment, etc.) are adopted to correct the coordinate conversion results. This process further improves the accuracy and reliability of three-dimensional space coordinates. The S23 technical solution not only improves the accuracy and efficiency of a single link through the optimization and collaborative work of multiple links, but also improves the overall performance of the entire system. This enables the system to maintain stable performance output when facing complex or changing environments, meeting the needs of various application scenarios.

[0110] In one embodiment of the present invention, S3 includes:

[0111] S31, using a path optimization algorithm based on a prediction model, predicting the future position according to the movement trend of the marker point, planning the optimal scanning path; and smoothing the planned path;

[0112] S32, based on the real-time data feedback mechanism, feeding back the latest position information of the LED tracking system to the scanning path planning system in real time;

[0113] S33. According to the real-time position information, the scanning path is dynamically adjusted so that the laser beam is always focused on the key part of the measured area.

[0114] The working principle of the above technical solution is as follows: using a path optimization algorithm based on a prediction model, the algorithm can analyze the historical motion trajectory and current motion state of the LED marker point, and predict its future motion trend and possible position through a mathematical model or a machine learning algorithm; based on the predicted future position information, the algorithm calculates an optimal scanning path from the current position to the expected target position. The determination of this optimal path takes into account a variety of factors, such as path length, scanning efficiency, obstacle avoidance, etc.; in order to ensure the smoothness and accuracy of the scanning process, the planned path is smoothed. This usually involves removing sharp turning points or unnecessary twists and turns in the path to make the path smoother and reduce shock and vibration in mechanical movement; the LED tracking system collects the latest position information of the LED marker point in real time, which is obtained through high-precision sensors and image processing technology to ensure the accuracy and real-time nature of the data; the collected position information is fed back to the scanning path planning system in real time through an efficient data transmission mechanism (such as wireless communication, wired connection, etc.). This step ensures that the system can respond quickly to changes in the external environment; after receiving the real-time position information, the scanning path planning system immediately performs data processing and analysis to evaluate the accuracy and effectiveness of the current path; based on the real-time position information and the results of the prediction model, the system evaluates the current scanning path. If it is found that the path deviates from the expected target or there is a better scanning strategy, the system will trigger the dynamic adjustment mechanism; based on the evaluation results, the system dynamically adjusts the scanning path. This includes changing the scanning direction, speed or path shape to ensure that the laser beam is always focused on the key part of the measured area. During the adjustment process, the system will consider a variety of constraints, such as mechanical motion restrictions, obstacle locations, etc.; the entire dynamic adjustment process forms a closed-loop control system. The system continuously receives real-time position information, evaluates the path effect, and makes dynamic adjustments accordingly. This closed-loop control mechanism ensures the stability and accuracy of the scanning process.

[0115] The effect of the above technical solution is: through the path optimization algorithm based on the prediction model, the system can predict the future position of the marking point in advance and plan the optimal scanning path accordingly. This forward-looking planning reduces unnecessary scanning actions and repeated scanning, thereby significantly improving the scanning efficiency. At the same time, the application of the prediction model also improves the accuracy of the scanning, because the system can more accurately locate the key part of the measured area; based on the real-time data feedback mechanism, the latest position information of the LED tracking system can be transmitted to the scanning path planning system in real time. This real-time performance ensures that the system can quickly respond to changes in the external environment, such as the sudden movement of the marking point or the appearance of new obstacles; based on the real-time position information, the system can dynamically adjust the scanning path. This dynamic adjustment capability enables the laser beam to always focus on the key part of the measured area, and maintain high-precision scanning even when the measured object or environment changes; smoothing the planned path helps to reduce shock and vibration in mechanical movement, thereby reducing the mechanical wear of the equipment. This not only extends the service life of the equipment, but also reduces the cost caused by frequent repairs and replacement of parts; by optimizing the scanning path and reducing unnecessary scanning actions, the system can use energy more efficiently and reduce overall energy consumption. This is especially important for large equipment that needs to run for a long time or in scenarios with limited energy. The efficient and accurate scanning process improves the user experience. Users no longer need to wait for too long for scanning or worry about the accuracy of the scanning results. The system's ability to respond to environmental changes in real time and dynamically adjust the scanning path makes it more flexible and adaptable. This helps to meet the needs of different users in different scenarios and improves user satisfaction. The efficiency and accuracy of the S3 technical solution enable it to be applied to more complex scenarios and tasks. For example, in the fields of industrial automation, robot navigation, 3D scanning, etc., this technical solution can play an important role. By introducing the S3 technical solution, enterprises or products can gain stronger competitiveness in the market. Its high efficiency, accuracy and flexibility make users more inclined to choose such products or services.

[0116] In one embodiment of the present invention, the S31 includes:

[0117] Obtain historical data and analyze and process the acquired historical data, including the historical position, speed, acceleration, etc. of the LED marker, and clean, denoise and normalize the data; select features (such as speed change rate, acceleration direction, trajectory curvature, etc.) according to the motion characteristics of the LED marker, extract the selected features, and construct the feature space of the prediction model;

[0118] Through deep learning algorithms (such as LSTM, GRU, Transformer, etc.), combined with historical data and feature space, a model that can predict the future position of LED markers is trained;

[0119] Using the trained prediction model, the future position of the LED marker is predicted, the prediction results of different time steps (such as short-term, medium-term, and long-term) are considered, and the uncertainty of the prediction results is evaluated; the reliability of the prediction results is quantified to provide a decision basis for subsequent path planning. Uncertainty assessment can be achieved by calculating the confidence interval and variance of the prediction results.

[0120] According to the specific needs of the application scenario, set the objective function of the scanning path, such as minimizing scanning time, maximizing scanning coverage, reducing energy consumption, etc. Consider various constraints in the actual scanning process, such as the maximum scanning speed of the laser, the minimum scanning distance, obstacle avoidance requirements, etc., to ensure that the planned path is both efficient and safe. Use path optimization algorithms (such as genetic algorithms, particle swarm optimization algorithms, etc.) combined with predicted future positions and uncertainty assessment results to plan the optimal scanning path;

[0121] Perform curvature analysis on the planned path to identify possible sharp change points or inflection points in the path; use smoothing algorithms (such as Bezier curve fitting, moving average filtering, etc.) to smooth the sharp change points in the path.

[0122] The working principle of the above technical solution is as follows: historical data of LED markers, including position, speed, acceleration and other information, are obtained from data sources (such as sensors, databases, etc.). Data is cleaned to remove outliers and noise to ensure data accuracy and reliability. Data is normalized so that data of different dimensions can be compared and analyzed on the same scale; according to the motion characteristics of LED markers, features that can reflect their motion trends, such as velocity change rate, acceleration direction, trajectory curvature, etc., are selected; the selected features are extracted to construct the feature space of the prediction model. These features will be used as input data for model training; deep learning algorithms (such as LSTM, GRU, Transformer, etc.) are used, which are good at processing time series data and can capture long-term dependencies in data; the prediction model is trained by combining historical data and feature space. During the training process, the model will learn how to predict the future position of the LED marker based on historical data; the future position of the LED marker is predicted using the trained prediction model. The prediction can cover different time steps (short-term, medium-term, and long-term) to meet the needs of different application scenarios; the prediction results are evaluated for uncertainty and the reliability of the prediction results is quantified. This can be achieved by calculating the confidence interval and variance of the prediction results. Uncertainty assessment helps to consider risk and uncertainty factors in subsequent path planning; according to the specific needs of the application scenario, set the objective function of the scanning path, such as minimizing scanning time, maximizing scanning coverage, reducing energy consumption, etc.; consider various constraints in the actual scanning process, such as the maximum scanning speed of the laser, the minimum scanning spacing, obstacle avoidance requirements, etc. These constraints will be used as limiting conditions for path planning; use path optimization algorithms (such as genetic algorithms, particle swarm optimization algorithms, etc.) to plan the optimal scanning path in combination with the predicted future position and uncertainty assessment results. The algorithm will comprehensively consider the objective function and constraints to find the best path that meets the conditions; perform curvature analysis on the planned path to identify possible sharp change points or inflection points in the path. These points may cause shock and vibration in mechanical movement, affecting the scanning effect and equipment life; use smoothing algorithms (such as Bezier curve fitting, moving average filtering, etc.) to smooth the sharp change points in the path. The smoothed path is smoother, which helps to reduce shock and vibration in mechanical movement and improve the stability and accuracy of scanning.

[0123] The effects of the above technical solutions are as follows: by cleaning, denoising and normalizing the historical data, the quality and consistency of the data are improved, laying a solid foundation for subsequent feature extraction and model training; according to the motion characteristics of the LED markers, key features are selected and extracted to construct the feature space of the prediction model. These features can more accurately reflect the motion trend of the markers, thereby improving the accuracy of the prediction; deep learning algorithms such as LSTM, GRU, and Transformer are used for model training, which can capture complex patterns and long-term dependencies in the data, further improving the accuracy of the prediction; uncertainty assessment is performed on the prediction results to quantify the reliability of the prediction results. This helps to consider risk and uncertainty factors in path planning and make more robust decisions; considering the prediction results of different time steps provides more comprehensive information for path planning. This helps the system make reasonable plans at different time scales and improve the flexibility and adaptability of decision-making; setting the objective function according to the specific needs of the application scenario, and considering various constraints in the actual scanning process. This ensures that the planned path is both efficient and safe and can meet the needs of actual applications; using path optimization algorithms such as genetic algorithms and particle swarm optimization algorithms, combined with the predicted future position and uncertainty assessment results, the optimal scanning path is planned. These algorithms can find the optimal or near-optimal solution in a complex environment, improve scanning efficiency and quality; perform curvature analysis on the planned path, identify possible sharp change points or inflection points in the path, and use a smoothing algorithm to process them. This helps reduce shock and vibration in mechanical movement and improve scanning stability and accuracy; smooth paths reduce wear and loss of mechanical parts, extend the service life of the equipment, and reduce maintenance costs; by optimizing the scanning path and reducing unnecessary scanning actions, scanning efficiency is improved and user waiting time is shortened; accurate prediction and path planning ensure that the laser beam can always focus on the key parts of the measured area, improving the accuracy and reliability of the scan; the system can respond to environmental changes in real time and dynamically adjust the scanning path, enhancing the flexibility and adaptability of the system and improving user satisfaction.

[0124] In one embodiment of the present invention, the S4 includes:

[0125] S41, according to the adjusted scanning path, start the 3D laser scanner to perform the scanning task, monitor the scanning process, and receive the returned scanning data;

[0126] S42, collecting three-dimensional coordinate data of all scanning points, wherein the three-dimensional coordinate data includes X, Y, Z coordinates and possible reflection intensity information;

[0127] S43. Perform quality checks on the collected raw data, remove noise points and outliers caused by occlusion, multipath reflection, etc.; filter and smooth the valid data; reduce random errors and improve data consistency and accuracy.

[0128] S44, and organize the processed three-dimensional coordinate data into a point cloud format to provide a basis for subsequent data processing and analysis.

[0129] The working principle of the above technical solution is as follows: according to the adjusted scanning path, the three-dimensional laser scanner is started to perform the scanning task. This step is the beginning of data acquisition, ensuring that the laser beam can be stably emitted according to the predetermined path; during the scanning process, the system monitors the emission state of the laser beam in real time to ensure the stability and accuracy of the laser beam. At the same time, the system also receives and records the returned scanning data, which is the basis for subsequent processing and analysis; the three-dimensional laser scanner measures the time difference (or phase difference) from the laser beam being emitted to being reflected by the object and then returning to the receiver, combined with the geometric parameters and scanning angle of the scanner, to calculate the three-dimensional coordinate data (X, Y, Z coordinates) of each scanning point. In addition, some scanners can also record reflection intensity information, which is helpful for subsequent data analysis and processing; the collected three-dimensional coordinate data is stored in the system's memory or in an external storage device for subsequent processing and analysis; the collected raw data is quality checked to eliminate noise points and outliers caused by factors such as occlusion and multi-path reflection. These noise points and outliers will interfere with the subsequent data processing and analysis results, so they must be eliminated; the valid data is filtered and smoothed to reduce random errors and improve the consistency and accuracy of the data. Filtering can remove high-frequency noise in the data, while smoothing can make the data smoother and reduce mutation points in the data; the processed 3D coordinate data is organized into a point cloud format. Point cloud is a commonly used 3D data representation method. It consists of a large number of points, each of which contains its 3D coordinate information (X, Y, Z) and possible reflection intensity information. The point cloud format facilitates subsequent data processing and analysis, such as 3D modeling, surface reconstruction, volume calculation, etc.; the organized point cloud data is output to a specified file or database for subsequent data processing and analysis.

[0130] The effect of the above technical solution is: by checking the quality of the collected raw data, the noise points and outliers caused by occlusion, multi-path reflection, etc. can be effectively eliminated. If these noises and outliers are not eliminated, they will have a negative impact on the subsequent data processing and analysis results, reducing the accuracy and reliability of the data; filtering and smoothing the valid data can further reduce random errors and improve the consistency and accuracy of the data. Filtering can remove high-frequency noise in the data, and smoothing can make the data smoother and reduce mutation points in the data, thereby improving the overall quality of the data; performing scanning tasks according to the adjusted scanning path can ensure that the scanning process is more efficient and orderly. This avoids invalid scanning and repeated scanning, improves scanning efficiency, and shortens scanning time; real-time monitoring of the scanning process ensures that the laser beam is stably emitted, which helps to promptly discover and solve problems that may arise during the scanning process and ensure the smooth progress of the scanning task; the processed three-dimensional coordinate data is organized into a point cloud format, providing a high-quality data foundation for subsequent data processing and analysis. Point cloud format is an intuitive and easy-to-process 3D data representation method, which is convenient for advanced data processing and analysis such as 3D modeling, surface reconstruction, and volume calculation. A high-quality data foundation can significantly improve the accuracy and efficiency of data processing and analysis. In the subsequent data processing and analysis process, useful information can be extracted more accurately, providing a more reliable basis for decision-making and application. Whether it is industrial automation, robot navigation, 3D reconstruction or other fields that require high-precision 3D data, the S4 technical solution can provide high-quality 3D coordinate data, thereby enhancing the practicality and effectiveness of these application scenarios. High-quality data and efficient scanning process can enhance the user experience. Users can obtain more accurate and reliable 3D data, while saving time and cost and improving work efficiency.

[0131] In one embodiment of the present invention, S5 includes:

[0132] S51. Extract common features from the marker point positions obtained by the LED optical tracking system and the point cloud obtained by the 3D laser scanner, such as edges, corner points, etc.

[0133] S52, using iterative closest point (ICP), accurately matching the position of the marker point with the corresponding feature in the point cloud; and calculating the transformation matrix from the marker point coordinate system to the point cloud coordinate system according to the matching result;

[0134] S53, aligning the three-dimensional point cloud data from different time points according to the timestamps, and fusing the point cloud data of adjacent time points through a sliding window according to the continuity and smoothness of the object's motion, to generate a continuous dynamic point cloud sequence;

[0135] S54. Use point cloud data to construct a surface geometric model of the object being measured through triangulation; dynamically adjust and update the three-dimensional model based on the real-time updated point cloud data to reflect the latest shape and motion status of the object.

[0136] S55. Optimize the constructed dynamic 3D model, such as smoothing, removing redundant faces, etc., to improve the visual effect and computational efficiency of the model. And display the dynamic 3D model through 3D visualization software or platform.

[0137] The working principle of the above technical solution is: extract common features such as edges and corners from the marker point positions obtained by the LED optical tracking system and the point cloud obtained by the 3D laser scanner. These common features are the basis for subsequent precise matching; describe the extracted features for matching in subsequent steps. Feature description usually includes information such as the position, shape, and size of the feature; use the iterative closest point (ICP) algorithm to accurately match the marker point position and the corresponding features in the point cloud. The ICP algorithm continuously searches for the optimal transformation matrix in an iterative manner to minimize the distance between the corresponding points of the two point sets; calculate the transformation matrix from the marker point coordinate system to the point cloud coordinate system based on the matching results. This transformation matrix describes the rotation and translation relationship between the two coordinate systems, and realizes the precise registration of the two; align the 3D point cloud data from different time points according to the timestamp to ensure the temporal consistency of the data. This is the basis for constructing a continuous dynamic point cloud sequence; according to the continuity and smoothness of the object's motion, the point cloud data of adjacent time points are fused through a sliding window. The size of the sliding window can be adjusted according to the actual situation to balance the computational efficiency and data accuracy; a continuous dynamic point cloud sequence is generated by continuously fusing the point cloud data of adjacent time points. This sequence reflects the shape and motion state of the object at different time points; the surface geometric model of the measured object is constructed by triangulation algorithm using the point cloud data. Triangulation is the process of converting point cloud data into a triangular mesh, which allows the surface of the object to be represented by a series of interconnected triangles; the 3D model is dynamically adjusted and updated according to the point cloud data updated in real time. This includes adding new triangles, deleting old triangles, and adjusting the position and shape of triangles to reflect the latest shape and motion state of the object; the constructed dynamic 3D model is optimized, such as smoothing and removing redundant facets. These optimization operations can improve the visual effect and computational efficiency of the model, making the model more realistic and easy to handle; the optimized dynamic 3D model is displayed through 3D visualization software or platform. This helps users intuitively understand and analyze the measurement results and improve the transparency and credibility of data processing.

[0138] The effect of the above technical solution is: by using the iterative closest point (ICP) algorithm, the position of the marker point obtained by the LED optical tracking system and the corresponding features in the point cloud obtained by the three-dimensional laser scanner are accurately matched, which can significantly improve the accuracy of data fusion. This high-precision matching lays a solid foundation for subsequent data processing and analysis; the transformation matrix calculated according to the matching results can accurately describe the rotation and translation relationship between the marker point coordinate system and the point cloud coordinate system, thereby achieving accurate registration of the two. This registration method can reduce the error caused by inconsistent coordinate systems and improve the accuracy and reliability of the overall data; the three-dimensional point cloud data from different time points are aligned according to the timestamp to ensure the time consistency of the data. This provides the necessary conditions for constructing a continuous dynamic point cloud sequence; by fusing the point cloud data of adjacent time points through a sliding window, the continuity and smoothness of the object's movement can be fully utilized to generate a more continuous and smooth dynamic point cloud sequence. This sequence not only reflects the morphological changes of the object at different time points, but also reveals its motion laws and dynamic characteristics; according to the point cloud data updated in real time, the three-dimensional model is dynamically adjusted and updated to reflect the latest morphology and motion state of the object. This real-time update capability enables the 3D model to always be consistent with the actual object, improving the real-time and accuracy of the measurement; using point cloud data to construct the surface geometry model of the measured object through triangulation can more intuitively display the 3D shape and structure of the object. This model is not only convenient for subsequent data processing and analysis, but also can provide strong support for the user's intuitive understanding and judgment; optimizing the constructed dynamic 3D model, such as smoothing and removing redundant faces, can significantly improve the visual effect and computational efficiency of the model. The optimized model is more realistic and concise, and can better meet the needs and expectations of users; displaying the optimized dynamic 3D model through 3D visualization software or platform can facilitate users to intuitively understand and analyze the measurement results. This visualization method not only improves the transparency and credibility of data processing, but also provides strong support for users' decision-making.

[0139] In one embodiment of the present invention, the S53 includes:

[0140] Through the timestamp alignment algorithm, data from different sources are arranged in chronological order to eliminate time deviations; based on the timestamp aligned data, preliminary data quality assessment is performed, such as checking point cloud density, noise level, etc. Low-quality or abnormal data frames are eliminated to ensure the accuracy and efficiency of subsequent processing.

[0141] Analyze the overall motion trajectory of the object using the marker point position information provided by the LED optical tracking system; predict the displacement and rotation trend of the object between adjacent time points through trajectory analysis;

[0142] Based on the results of motion trajectory analysis, the point cloud data of adjacent time points are roughly aligned; simple transformations (such as translation and rotation) are used to roughly align the point cloud data to the same position;

[0143] The sliding window size is set according to the requirements of the application scenario and the characteristics of the point cloud data. The window size should be able to contain enough point cloud data to reflect the motion state of the object while avoiding excessive computational burden. Within the sliding window, a point cloud fusion algorithm (such as fusion based on statistical information, fusion based on geometric features, etc.) is used to fuse the point cloud data of adjacent time points. The continuity and smoothness of the object's motion are considered during the fusion process to ensure that the fused point cloud data is both accurate and continuous.

[0144] Perform a quality assessment on the fused point cloud data to check whether the fusion effect meets the requirements. The evaluation indicators may include point cloud density change, shape consistency, noise level, etc. Connect the fused point cloud data in the sliding window in chronological order to construct a continuous dynamic point cloud sequence. Each frame in the sequence represents the 3D shape and motion state of the object at a certain moment.

[0145] The working principle of the above technical solution is as follows: through the timestamp alignment algorithm, the data from different sources (such as scanning at different time points or acquisition by different sensors) are arranged in chronological order. This step is to eliminate the time deviation and ensure the consistency of the data in time; then, the data after timestamp alignment is preliminarily evaluated for quality. This includes checking key indicators such as point cloud density and noise level to eliminate low-quality or abnormal data frames. This step is an important prerequisite for ensuring the accuracy and efficiency of subsequent processing; the overall motion trajectory of the object is analyzed using the marker point position information provided by the LED optical tracking system. Through trajectory analysis, the displacement and rotation trends of objects between adjacent time points can be predicted to provide guidance for subsequent precise registration; based on the results of motion trajectory analysis, the point cloud data at adjacent time points are roughly registered. This step uses simple transformations (such as translation and rotation) to roughly align the point cloud data to the same position to reduce the amount of calculation for subsequent precise registration; the size of the sliding window is set according to the needs of the application scenario and the characteristics of the point cloud data. The window size should be selected to include enough point cloud data to reflect the motion state of the object while avoiding excessive computational burden; within the sliding window, the point cloud fusion algorithm is used to fuse the point cloud data of adjacent time points. The fusion algorithm can be based on statistical information (such as weighted average), geometric features (such as feature matching) and other methods. In the fusion process, it is necessary to fully consider the continuity and smoothness of the object's motion to ensure that the fused point cloud data is both accurate and continuous; the quality of the fused point cloud data is evaluated to check whether the fusion effect meets the requirements. The evaluation indicators may include point cloud density changes, shape consistency, noise level, etc. This step is to ensure that the quality of the fused data meets the standards for subsequent processing and analysis; finally, the fused point cloud data in the sliding window are connected in chronological order to construct a continuous dynamic point cloud sequence. Each frame in the sequence represents the three-dimensional shape and motion state of the object at a certain moment, providing a rich data source for subsequent data processing, analysis and visualization.

[0146] The effects of the above technical solution are: through the timestamp alignment algorithm, the temporal consistency of data from different sources is ensured, the time deviation is eliminated, and an accurate time reference is provided for subsequent data processing; the preliminary data quality assessment can timely discover and eliminate low-quality or abnormal data frames, avoiding the negative impact of these data on the subsequent processing results, thereby improving the accuracy and efficiency of the overall data processing; the marker point position information provided by the LED optical tracking system can accurately analyze the overall motion trajectory of the object. This analysis not only helps to understand the motion law of the object, but also provides strong support for the subsequent rough alignment and precise alignment; the displacement and rotation trend of the object between adjacent time points is predicted through trajectory analysis, which provides important reference information for the subsequent rough alignment and point cloud fusion, and helps to achieve more accurate data alignment and fusion; the rough alignment based on the motion trajectory analysis results can roughly align the point cloud data of adjacent time points to the same position, reducing the calculation amount of subsequent precise alignment and improving the efficiency of data processing; the sliding window size is set according to the needs of the application scenario and the characteristics of the point cloud data, and the appropriate point cloud fusion algorithm is used in the window to fuse the point cloud data of adjacent time points. This fusion method not only takes into account the continuity and smoothness of the object's motion, but also avoids excessive computational burden, achieving efficient data fusion; the quality of the fused point cloud data is evaluated to ensure that the fusion effect meets the requirements. This evaluation mechanism helps to timely discover and correct problems that may occur in the fusion process, ensuring the quality of the final constructed dynamic point cloud sequence; the fused point cloud data in the sliding window are connected in chronological order to construct a continuous dynamic point cloud sequence. This sequence not only reflects the three-dimensional shape and motion state of the object at different time points, but also provides a rich data source for subsequent data processing, analysis and visualization.

[0147] One embodiment of the present invention, as Figure 2 As shown, a three-dimensional laser scanner measurement system, the system comprises:

[0148] Initialization setting module: initializing the device, including arranging multiple LED marking points on the object to be measured, starting the LED optical tracking system, and setting the parameters of the 3D laser scanner;

[0149] Data processing module: The LED optical tracking system obtains the video stream containing the LED marker points in real time, transmits the video stream to the cloud space for real-time processing, identifies and extracts the position information of each LED marker point, and calculates its motion trajectory and speed according to the position change of the marker point;

[0150] Path update module: dynamically adjust the scanning path of the 3D laser scanner based on the position and motion trajectory of the LED marker point; specifically, plan the scanning path as a path around or following the marker point to ensure that the laser beam can fully cover the measured area and reduce invalid scanning; and update the scanning path in real time based on the marker point position information obtained in real time to cope with the dynamic changes of the measured object;

[0151] Point cloud generation module: Start the 3D laser scanner, scan according to the planned scanning path, and collect the 3D coordinate data of all scanning points to form a 3D point cloud of the surface of the object being measured;

[0152] Model generation module: align the marker point position information obtained by the LED optical tracking system with the point cloud data obtained by the 3D laser scanner; establish a mapping relationship between the two by calculating the corresponding positions of the marker points in the 3D point cloud; combine the 3D point cloud data at different time points with the timestamp information to form a dynamic 3D model of the object being measured.

[0153] 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 claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A three-dimensional laser scanner measurement method, characterized in that: The method comprises: S1, initializing the device, wherein the initialization setting includes arranging a plurality of LED marking points on the object to be measured, starting the LED optical tracking system, and setting the parameters of the three-dimensional laser scanner; S2, according to the LED optical tracking system, real-time acquisition of the video stream containing the LED marking points, and transmission of the video stream to the cloud space for real-time processing, identification and extraction of the position information of each LED marking point, and calculation of its motion trajectory and speed according to the position change of the marking point; S3, dynamically adjust the scanning path of the 3D laser scanner based on the position and motion trajectory of the LED marker point; and update the scanning path in real time according to the marker point position information obtained in real time; S4, start the 3D laser scanner, scan according to the planned scanning path, and collect the 3D coordinate data of all scanning points to form a 3D point cloud of the surface of the object to be measured; S5. Align the marker point position information obtained by the LED optical tracking system with the point cloud data obtained by the 3D laser scanner; establish a mapping relationship between the two by calculating the corresponding position of the marker point in the 3D point cloud; combine the 3D point cloud data at different time points with the timestamp information to form a dynamic 3D model of the object being measured.

2. The three-dimensional laser scanner measurement method according to claim 1, characterized in that: Said S1 comprises: S11. According to the shape, size and expected range of motion of the object to be measured, the distribution of the marking points is simulated by computer simulation software, and the LED marking points are fixed at the key positions of the object to be measured by a high-precision mechanical arm; S12, uniquely number each LED marking point and record its initial position information through a static calibration procedure; S13, after starting the system, perform self-test and preheating at the same time; adjust the parameters of the camera according to the size, moving speed and ambient light conditions of the object to be measured; S14. Select the scanning accuracy and scanning speed according to the measurement requirements, and balance the measurement accuracy and measurement time; set the scanning range and scanning angle of the laser scanner according to the size and shape of the object to be measured; and configure the data transmission protocol and storage path.

3. The three-dimensional laser scanner measurement method according to claim 1, characterized in that: The S2 comprises: S21, continuously capture the video stream containing the LED marker points through a high frame rate camera; perform real-time compression processing on the video stream and transmit it to the cloud space through wired or wireless means; S22, the cloud space pre-processes the received video frames and identifies the LED marking points in the video through a machine learning algorithm; S23, based on the recognition result, extract the pixel coordinates of each marking point, and convert them into three-dimensional space coordinates through camera calibration parameters; S24. According to the position information of the marker point in the continuous frames, the Kalman filter algorithm is used to track its motion trajectory; based on the trajectory data, the instantaneous velocity and acceleration of the marker point are calculated.

4. The three-dimensional laser scanner measurement method according to claim 3, characterized in that: The S21 includes: The video stream containing the LED markers is captured by a high-frequency camera. During the video capture process, the video stream is previewed in real time, and the camera parameters are dynamically adjusted according to the preview effect. Through the video encoding algorithm, the captured video stream is efficiently compressed. According to the dynamic changes of the video content, based on the preset intelligent compression strategy, the key frames and key areas are encoded with high quality, and the non-key frames and background areas are compressed with low quality. During the compression process, the video frames are preprocessed by real-time denoising and contrast enhancement algorithms; According to the on-site environment and network conditions, select a combination of wired and wireless transmission methods to transmit the video stream; Using multi-path transmission technology, the video stream is divided into multiple data packets and transmitted in parallel to the cloud space through different network paths.

5. The three-dimensional laser scanner measurement method according to claim 3, characterized in that: The S23 comprises: Based on the recognition of LED markers by machine learning algorithms, the edges of the markers are further refined through morphological operations; Through sub-pixel positioning technology, pixel coordinates are extracted on the refined edges; Through the camera calibration method, combined with the standard calibration plate or three-dimensional calibration target, the camera's intrinsic parameter matrix and extrinsic parameter matrix are obtained; According to the calibration results, the image in the video frame is distorted and corrected, and the perspective projection transformation matrix from the pixel coordinate system to the camera coordinate system is constructed based on the pinhole camera model; Combined with the position and posture of the camera in the world coordinate system, the coordinates in the camera coordinate system are converted to the world coordinate system; If the LED marker itself does not contain depth information, multi-view, structured light or ToF sensor is used to assist in depth estimation; The depth estimation results are combined with the perspective projection model to optimize the calculation of three-dimensional space coordinates, and the various error sources that affect the accuracy of coordinate transformation are analyzed. Based on the identified error sources, the corresponding correction method is used to correct the coordinate transformation results.

6. The three-dimensional laser scanner measurement method according to claim 1, characterized in that: The S3 includes: S31, using a path optimization algorithm based on a prediction model, predicting the future position according to the movement trend of the marker point, planning the optimal scanning path; and smoothing the planned path; S32, based on the real-time data feedback mechanism, feeding back the latest position information of the LED tracking system to the scanning path planning system in real time; S33. According to the real-time position information, the scanning path is dynamically adjusted so that the laser beam is always focused on the key part of the measured area.

7. The three-dimensional laser scanner measurement method according to claim 1, characterized in that: The S3 includes: Obtain historical data, analyze and process the acquired historical data, select features based on the motion characteristics of the LED markers, extract the selected features, and construct the feature space of the prediction model; Through deep learning algorithms, historical data and feature space are combined to train a model that can predict the future position of LED markers; Use the trained prediction model to predict the future position of the LED marker and perform uncertainty assessment on the prediction results; According to the specific needs of the application scenario, the objective function of the scanning path is set, and the optimal scanning path is planned through the path optimization algorithm combined with the predicted future position and uncertainty assessment results; Perform curvature analysis on the planned path to identify possible sharp change points or inflection points in the path; use a smoothing algorithm to smooth the sharp change points in the path.

8. The three-dimensional laser scanner measurement method according to claim 1, characterized in that: The S4 comprises: S41, according to the adjusted scanning path, start the 3D laser scanner to perform the scanning task, monitor the scanning process, and receive the returned scanning data; S42, collecting three-dimensional coordinate data of all scanning points, wherein the three-dimensional coordinate data includes X, Y, Z coordinates and reflection intensity information; S43, performing quality check on the collected raw data, removing noise points and outliers; filtering and smoothing the valid data; S44, and organize the processed three-dimensional coordinate data into a point cloud format.

9. The three-dimensional laser scanner measurement method according to claim 1, characterized in that: The S5 comprises: S51, extracting common features from the marker point positions obtained by the LED optical tracking system and the point cloud obtained by the three-dimensional laser scanner; S52, using iterative nearest point, matching the position of the marker point with the corresponding feature in the point cloud; and calculating the transformation matrix from the marker point coordinate system to the point cloud coordinate system according to the matching result; S53, aligning the three-dimensional point cloud data from different time points according to the timestamps, and fusing the point cloud data of adjacent time points through a sliding window according to the continuity and smoothness of the object's motion, to generate a continuous dynamic point cloud sequence; S54, using the point cloud data to construct a surface geometric model of the object being measured through triangulation; dynamically adjusting and updating the three-dimensional model according to the point cloud data updated in real time; S55. Optimize the constructed dynamic three-dimensional model and display the dynamic three-dimensional model through three-dimensional visualization software or platform.

10. A three-dimensional laser scanner measurement system, characterized in that: The system comprises: Initialization setting module: initializing the device, including arranging multiple LED marking points on the object to be measured, starting the LED optical tracking system, and setting the parameters of the 3D laser scanner; Data processing module: The LED optical tracking system obtains the video stream containing the LED marker points in real time, transmits the video stream to the cloud space for real-time processing, identifies and extracts the position information of each LED marker point, and calculates its motion trajectory and speed according to the position change of the marker point; Path update module: dynamically adjust the scanning path of the 3D laser scanner based on the position and motion trajectory of the LED marker points; and update the scanning path in real time based on the marker point position information obtained in real time; Point cloud generation module: Start the 3D laser scanner, scan according to the planned scanning path, and collect the 3D coordinate data of all scanning points to form a 3D point cloud of the surface of the object being measured; Model generation module: align the marker point position information obtained by the LED optical tracking system with the point cloud data obtained by the 3D laser scanner; establish a mapping relationship between the two by calculating the corresponding positions of the marker points in the 3D point cloud; combine the 3D point cloud data at different time points with the timestamp information to form a dynamic 3D model of the object being measured.

Images