Remote positioning system for nuclear reactor internals in three-dimensional space
Through binocular structured light camera and image processing technology, combined with Yolov5 and semantic segmentation model, the problems of low efficiency and safety hazards in traditional positioning methods in nuclear reactors are solved, and high-precision unmanned remote operation and three-dimensional spatial positioning are achieved, which protects the Workers are safe.
Patent Information
- Application Number
- CN202410924372.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-06-24
- Filing Date
- 2024-07-11
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In nuclear reactors, the traditional artificial positioning method is inefficient and has radiation safety risks, and there is no effective binocular vision industry ultra-long-distance three-dimensional spatial positioning detection method.
Using binocular structured light camera combined with Yolov5 object detection model, KNN algorithm, ESRGAN super-resolution algorithm and semantic segmentation model, the high-precision three-dimensional spatial positioning of components in the nuclear reactor through image processing technology, and using HSV color threshold detection and binocular camera imaging Principle calculates three-dimensional coordinates.
It realizes unmanned remote operation of high-precision guide positioning of components in nuclear industrial reactors, reduces labor intensity, improves positioning and installation efficiency, protects staff safety, and achieves high-precision positioning in water wading states.
Smart Images

Figure CN119006590A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of remote control, and in particular to a remote positioning system for nuclear reactor internal components in three-dimensional space. Background Art
[0002] The reactor's lower internals are circular objects with a diameter of 6m. When positioning and installing the internals, the traditional manual positioning method is inefficient, and the presence of radiation is a serious safety hazard for workers. The present invention realizes the rapid positioning and installation of the maintenance and replacement of the internals, realizes the unmanned remote operation of high-precision guided positioning measurement, realizes the high-precision guided positioning of the measured object in the partial wading state, and outputs the coordinate position to facilitate the real-time positioning of the lower internals in space, and can assist the lifting tool to guide the lower internals into the core. At present, there is no related technology for industrial ultra-long-distance three-dimensional spatial positioning detection methods based on binocular vision. Summary of the invention
[0003] The object of the present invention is to provide a remote positioning system for nuclear reactor internal components in three-dimensional space to solve the problems raised in the above-mentioned background technology.
[0004] To achieve the above object, the present invention provides the following technical solutions:
[0005] A remote positioning system for nuclear reactor internals in three-dimensional space, comprising:
[0006] In a first aspect, a method for remotely positioning a nuclear reactor internal component in three-dimensional space comprises:
[0007] The target detection dataset is created by collecting color image information of nuclear reactor internal components at different angles, scenes and distances;
[0008] Using the Yolov5 target detection model, we set the network structure parameters that match the computing power of the image information processor and then trained the target detection model for reactor internal components.
[0009] The target detection result area is cropped and the cropped area is used as the source image for subsequent algorithms;
[0010] According to the target detection results of the color image, the corresponding area in the depth image is cropped and processed as the source image of the subsequent algorithm;
[0011] The KNN algorithm is used to complete the areas where depth information is missing in the depth map. In the KNN algorithm, the pixels where depth information is missing are first marked, and then the distance from the points where the depth value is missing to other points where the depth information is not missing is calculated. The calculation formula is as follows:
[0012]
[0013] in Represents a point (x (i) ,y (i) ) to point (x (j) ,y (j) ) distance.
[0014] The calculated distances are sorted, the K points with the closest distances are selected, and the KNN algorithm is used to determine the depth value of the missing pixel by weighted averaging or voting the depth values of the adjacent pixels.
[0015] Develop a plan for super-resolution image magnification. When the image size after target detection is less than 128×80, the magnification is 10 times; when the image size after target detection is less than 160*100, the magnification is 8 times; when the image size after target detection is less than 256*160, the magnification is 5 times; when the image size after target detection is less than 320*200, the magnification is 4 times; when the image size after target detection is less than 1640*400, the magnification is 2 times;
[0016] Using the same magnification to process the color image and the depth image can enrich the image details and improve the positioning detection accuracy for the color image; for the depth image, it can ensure that the depth information of the corresponding position in the depth image can match the color image;
[0017] Perform HSV color threshold detection and set the upper and lower limits of the threshold detection;
[0018] Convert the image to HSV color space;
[0019] Perform edge detection and contour screening on the image after HSV color space conversion;
[0020] Calculate the coordinates of feature points based on the selected contours and display the image and output the pixel coordinates of the feature points;
[0021] The pixel coordinates of the feature point detection output are restored. After the target is detected and clipped, the pixel coordinates of the reactor internal components change, but the relative positions and size ratios of the various parts of the reactor internal components remain unchanged. The pixel coordinates of the feature points after clipping and the pixel coordinates of the feature points before clipping have the following relationship:
[0022]
[0023] Among them, (x, y) represents the coordinates of the center point of the cropped image, (x1, y1) represents the pixel coordinates of the feature point detection output, (x2, y2) represents the pixel coordinates of the feature point detection output in the image before target detection cropping, (x0, y0) represents the pixel coordinates of the center point of the target detection cropping area, weight and height represent the width and length of the image after target detection cropping, weight1 and height1 represent the width and length of the image after super-resolution processing;
[0024] According to the imaging principle of binocular cameras, the pixel coordinate system and the world coordinate system have the following relationship:
[0025]
[0026] Among them, Z c Represents a point in the pixel coordinate system, with the origin in the upper left corner of the image, u to the right, and v to the bottom; represents the binocular internal parameter matrix, f x , f y Respectively represent the length of the focal length on the x-axis and y-axis, u0 and v0 are the optical centers of the image; [R 3×3 T 3×1 ] represents the binocular rotation matrix and translation matrix; (X W , Y W , Z W ) is the coordinate in the world coordinate system;
[0027] Read the pixel coordinates p(x,y) output by the feature point extraction part, then read the depth information of the corresponding position of the pixel coordinates in the depth map, and finally calculate the three-dimensional space coordinates of the pixel coordinates based on the imaging principle formula of the binocular camera;
[0028] On the basis of the image information collected when making the target detection dataset, new image information is collected, and more refined image annotation operations are performed on the expanded image information to make a semantic segmentation dataset;
[0029] The original images and label images in the dataset were mirrored left and right, mirrored up and down, and rotated 90°, 180°, and 270° clockwise. Finally, the dataset was expanded to six times its original size.
[0030] Use the Deeplabv3+ semantic segmentation model, set the network structure parameters that match the image information processor's computing power, and then train the semantic segmentation model for reactor internal components;
[0031] According to the pixel-level classification results of semantic segmentation, the corresponding positions in the color image and depth image are cropped, and the three-dimensional spatial coordinates corresponding to each point in the image are calculated according to the imaging principle of the binocular camera. They are then displayed in three dimensions to generate a three-dimensional reconstructed image of the internal components of the nuclear reactor.
[0032] Compared with the prior art, the present invention has the following beneficial effects:
[0033] The present invention utilizes a binocular structured light camera to collect information about ultra-long-distance target objects in industry, and uses image processing technologies such as target detection, KNN completion, ESRGAN super-resolution algorithm, semantic segmentation, and HSV thresholding to locate the pixel coordinate position of the target object, and then calculates the three-dimensional spatial coordinates of the target object based on the imaging principle of the binocular camera.
[0034] It fills the gap of ultra-long-distance binocular vision positioning in industry, realizes unmanned remote operation of high-precision guidance and positioning of nuclear industry reactor internal components, and realizes high-precision guidance and positioning of measured objects in partial wading state, which greatly reduces labor intensity and improves positioning and installation efficiency. It has the following advantages:
[0035] (1) The image information of ultra-long-distance target positioning objects in the nuclear industry is collected using a binocular structured light camera, and the accuracy of the information is better than that collected by a traditional binocular camera.
[0036] (2) Use target detection to reduce the image processing area, eliminate useless information collected, and concentrate computer computing power on the area where the target object is located, thereby reducing the amount of image processing tasks and improving the efficiency of target positioning detection.
[0037] (3) The ESRGAN super-resolution upscaling algorithm uses an adaptive magnification factor based on the size of the target detection result area to prevent information loss and enrich image details, which can effectively make up for the defect of low resolution of the binocular camera.
[0038] (4) Semantic segmentation is combined with a 3D point cloud reconstruction algorithm and a 3D point cloud registration algorithm to generate a 3D real-time animation of the target object, which facilitates all-round observation of the target object's posture and position in a large scene.
[0039] (5) It realizes unmanned remote operation and high-precision guidance and positioning of nuclear reactor internal components, effectively protecting the safety of workers during the installation of nuclear reactor internal components.
[0040] (6) It realizes high-precision guidance and positioning of the measured object in the partial wading state, with high degree of automation and high efficiency, which can greatly save labor costs. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1It is a work flow chart of the present invention;
[0042] Figure 2 This is a schematic diagram of the installation of reactor internal components;
[0043] Figure 3 is a target detection flow chart in an embodiment;
[0044] Figure 4 This is a diagram of the target detection network structure based on Yolov5 in the embodiment;
[0045] Figure 5 is a flow chart of super-resolution optimization in an embodiment;
[0046] Figure 6 is a flow chart of HSV-based threshold feature point detection in an embodiment;
[0047] Figure 7 is a schematic diagram of calculating three-dimensional space coordinates by pixel coordinates in Example 2;
[0048] Figure 8 It is a system block diagram of a nuclear reactor internal component remote positioning system based on binocular vision in an embodiment. DETAILED DESCRIPTION
[0049] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. 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.
[0050] Example 1, reference Figure 8 , a remote positioning system for nuclear reactor internals in three-dimensional space, characterized by comprising:
[0051] A hardware system module is used to obtain color image information and depth image information of the components in the pile and load the execution algorithm system module;
[0052] An algorithm system module, used for processing the color image information and the depth image information of the in-pile components to output a three-dimensional calculation and analysis result;
[0053] Software system module, used to display three-dimensional calculation and analysis results and user interaction.
[0054] Embodiment 2, based on embodiment 1, the hardware system module of the present invention includes: an image information collector, a digital information processor, an image information processor and a data storage device; wherein, the image information collector uses a binocular structured light camera, and the digital information processor, the image information processor and the data storage device use a computer.
[0055] Embodiment 3, based on embodiment 2, the algorithm system module of the present invention includes: a model building module, a KNN module, a super-resolution enlargement module and a three-dimensional space information calculation module.
[0056] Embodiment 4, on the basis of embodiment 3, the model construction module of the present invention constructs a three-dimensional space remote positioning device network model based on Yolov5 and Deeplabv3+ neural networks, and constructs a training data set of the three-dimensional space remote positioning device network model based on the in-pile components; the training data set includes the image data of the in-pile components at various distances and angles in the positioning site;
[0057] Among them, the KNN module is used to fill the missing part of the depth information in the depth map;
[0058] The super-resolution enlargement module is used to enrich the color image, increase the color vividness of the feature points, and make the depth information of the corresponding position in the depth image match the color image;
[0059] Three-dimensional spatial information calculation module, used for three-dimensional calculation analysis and three-dimensional point cloud reconstruction.
[0060] Embodiment 5, based on embodiment 4, the model building module of the present invention comprises:
[0061] The first network construction unit, based on the target detection results, synchronously crops the color image and the depth image as the input image of the KNN module and the super-resolution enlargement module;
[0062] The second network construction unit simultaneously crops the color image and the depth image based on the semantic segmentation result as the input image of the three-dimensional spatial information calculation module.
[0063] Example 6. Based on Example 5, the software system module of the present invention includes a graphical user interface which is a software interface of a multi-phase array visual positioning system, and the interface is used for image display, configuration parameters, control and operation of the multi-phase array visual positioning system; wherein the home page includes a navigation bar at the top, function buttons below the navigation bar, a camera color image display area below the function buttons, a data display area below the camera color image display area, and a picture display area and a three-dimensional model display area on the right side of the home page.
[0064] Embodiment 7, based on embodiment 6, the system software interface of the present invention further includes:
[0065] Click the "Deformation Calculator" function button in the main view to enter the deformation calculator page. At the top of the deformation calculator page are the deformation calculation formula 1 description area, calculation input area and calculation result display area. The deformation calculation formula 1 is:
[0066]
[0067] Among them, ΔL represents the deformation of the object (unit: meter, m), F represents the force applied to the object (unit: Newton, N), L represents the length of the object (unit: meter, m), Y represents the Young's modulus of the object (unit: Pascal, Pa), A represents the cross-sectional area of the object (unit: square meter, m 2 );
[0068] Below is the deformation calculation formula 2 description area, calculation input area and calculation result display area. The deformation calculation formula 2 is:
[0069]
[0070] Among them, ΔL represents the deformation of the object (unit: meter, m), σ represents the normal stress applied to the object (unit: Pascal, Pa), L represents the length of the object (unit: meter, m), and Y represents the Young's modulus of the object (unit: Pascal, Pa).
[0071] Embodiment 8, based on embodiment 7, the software system module of the present invention has the functions of three-dimensional analysis and calculation, three-dimensional visualization, object deformation calculation, receiving and real-time display of image data from the camera, and also includes:
[0072] Image display function: receiving an original image at a first preset time interval, the original image being taken by a binocular camera shooting unit in the nuclear reactor, and displaying the original image on the software;
[0073] Positioning point distance measurement function: Generate a three-dimensional point cloud image from the color image and the depth image, identify the positioning points in the three-dimensional point cloud image, calculate the distance between two positioning points through the coordinates of the positioning points in the three-dimensional point cloud image, and display the positioning point distance in the software;
[0074] 3D point cloud reconstruction image display function: read and display the 3D point cloud reconstruction image of the positioning bracket in the software at every second preset time interval;
[0075] Target object trajectory point calculation and display function: calculate the center coordinates of the positioning bracket as the positioning bracket trajectory point, and save the trajectory point and display it in the software;
[0076] 3D reconstruction function of the scene where the target object is located: obtain the color image and depth image of the scene where the positioning bracket is located, generate and save the 3D point cloud map of the scene through the color image and depth image;
[0077] Shooting and timed shooting function: take color pictures and depth pictures through the binocular camera and save them; take color pictures and depth pictures and save them by setting the shooting interval through the software;
[0078] Positioning point ranging function: control the binocular camera to capture color images and depth images and generate a 3D point cloud map, identify the positioning points, perform the positioning point ranging function and display the results in the software;
[0079] Image display function: read the image in the disk and display it in the software;
[0080] Format conversion function: read the raw format file, convert the raw format file into a png format file and save it;
[0081] 3D point cloud image generation function: read the color image and depth image taken by the binocular camera, generate a 3D point cloud image through the captured color image and depth image and save it;
[0082] 3D point cloud point number statistics and free point selection distance measurement function: read the 3D point cloud, calculate the number of points in the 3D point cloud and display it; select any two points in the 3D point cloud by mouse to measure the distance and display it;
[0083] Object deformation calculation function: by inputting the weight, length, Young's modulus and cross-sectional area of the object to be measured, the deformation of the object to be measured is calculated according to the given formula; by inputting the normal stress, length and Young's modulus of the object to be measured, the deformation of the object to be measured is calculated according to the given formula;
[0084] Binocular camera depth map working mode selection function: select the depth map working mode of the binocular camera; there are three working modes: precision enhancement mode, long-distance enhancement mode, and frame rate enhancement mode.
[0085] The working principle of the present invention is as follows:
[0086] The workflow diagram of the present invention is as follows Figure 1 shown.
[0087] During the process of positioning and installing the internal components, local wading may occur. At this time, the image information collector cannot collect complete image information of the internal components. Therefore, the present invention positions the positioning bracket model that has a rigid connection relationship with the internal components. The connection relationship between the internal components and the positioning bracket is shown in the figure. Figure 2 As shown, the purpose of indirectly positioning the components inside the pile can be achieved by positioning the positioning bracket.
[0088] Use a binocular structured light camera to scan and photograph the positioning bracket to obtain color image information and depth image information; the color image information and depth image information have the same resolution, and the color image information and depth image information stored at the same pixel points represent the same point in three-dimensional space; this must be guaranteed in subsequent algorithm processing.
[0089] Based on the Yolov5 target detection model, the color image is processed for target detection. The target detection flow chart is as follows: Figure 3 As shown; according to the flow chart, firstly, a target detection data set is prepared, and the data set size of the present invention is 813; according to the computing power of the image information processor in the present invention, the network structure parameters are set to obtain the target detection model for the positioning bracket through training, and the Yolov5 network structure is as follows Figure 4 As shown; in order to reduce the amount of calculation of the image information processor, the present invention uses the target detection result area as a reference to synchronously crop the depth image and the color image. In order to ensure that the image information of the positioning bracket is not lost during the cropping process, the length and width of the cropping area of the present invention are expanded by 20 pixels based on the target cropping result.
[0090] In the depth image collected by the graphic information collector, there will inevitably be areas where depth information is missing, which seriously affects the accuracy of the three-dimensional calculation and analysis part of the present invention. To address this problem, the present invention uses the KNN algorithm to fill in the areas where depth information is missing;
[0091] In this embodiment, when the positioning bracket runs to a position far away from the image information collector, the color image collected by the image information collector will appear blurred, the object boundaries are not clear, and the color vividness is low; to address this problem, a super-resolution amplification operation is performed on the color image to improve the image resolution, enrich the image details, and increase the image color vividness.
[0092] The traditional super-resolution magnification algorithm needs to manually set the magnification before each operation, and the degree of automation is too low, which seriously affects the real-time positioning function of the present invention. To address this problem, the present invention innovatively uses the ESRGAN super-resolution magnification algorithm with an adaptive magnification on the basis of the traditional super-resolution magnification algorithm. The ESRGAN super-resolution magnification algorithm with an adaptive magnification automatically sets the magnification according to the size of the image cropped by the target detection module: when the image size after target detection is less than 128×80, the magnification is 10 times; when the image size after target detection is less than 160*100, the magnification is 8 times; when the image size after target detection is less than 256*160, the magnification is 5 times; when the image size after target detection is less than 320*200, the magnification is 4 times; when the image size after target detection is less than 1640*400, the magnification is 2 times; the depth image magnification is consistent with the color image magnification, and the color image super-resolution magnification effect comparison diagram before and after Figure 5 shown.
[0093] The HSV threshold detection method is used to detect feature points on the super-resolution amplified color image to obtain the pixel coordinates of the positioning bracket. The HSV threshold detection flow chart is as follows: Figure 6 shown.
[0094] The pixel coordinates of the feature point detection output are restored. After the target is detected and clipped, the pixel coordinates of the reactor internal component positioning bracket are changed, but the relative positions and size ratios of the various parts of the reactor internal component positioning bracket object remain unchanged. The pixel coordinates of the feature points after clipping and the pixel coordinates of the feature points before clipping have the following relationship:
[0095]
[0096] Among them, (x, y) represents the coordinates of the center point of the cropped image, (x1, y1) represents the pixel coordinates of the feature point detection output, (x2, y2) represents the pixel coordinates of the feature point detection output in the image before target detection cropping, (x0, y0) represents the pixel coordinates of the center point of the target detection cropping area, weight and height represent the width and length of the image after target detection cropping, and weight1 and height1 represent the width and length of the image after super-resolution processing.
[0097] like Figure 7 As shown in the figure, according to the imaging principle of the binocular camera, the conversion relationship between the three-dimensional coordinates of a point in space and its corresponding pixel coordinates is as follows:
[0098]
[0099] Among them, Zc Represents a point in the pixel coordinate system, with the origin in the upper left corner of the image, u to the right, and v to the bottom; represents the binocular internal parameter matrix, f x , f y Respectively represent the length of the focal length on the x-axis and y-axis, u0 and v0 are the optical centers of the image; [R 3×3 T 3×1 ] represents the binocular rotation matrix and translation matrix; (X W , Y W , Z W ) are the coordinates in the world coordinate system.
[0100] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be considered exemplary and non-restrictive in all respects, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims be included in the present invention.
[0101] In addition, it should be understood that although this specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment have also been appropriately combined to form other implementation modes that are easy for those skilled in the art to understand.
Claims
1. A remote positioning system for nuclear reactor internals in three-dimensional space, characterized in that: include: A hardware system module is used to obtain color image information and depth image information of the components in the pile and load the execution algorithm system module; An algorithm system module, used for processing the color image information and the depth image information of the in-pile components to output a three-dimensional calculation and analysis result; Software system module, used to display three-dimensional calculation and analysis results and user interaction.
2. A system for remotely positioning nuclear reactor internals in three-dimensional space according to claim 1, characterized in that: The hardware system module includes: an image information collector, a digital information processor, an image information processor and a data storage device; wherein the image information collector uses a binocular structured light camera, and the digital information processor, the image information processor and the data storage device use a computer.
3. A system for remotely positioning nuclear reactor internals in three-dimensional space according to claim 1, characterized in that: The algorithm system module includes: a model building module, a KNN module, a super-resolution enlargement module and a three-dimensional space information calculation module.
4. A system for remotely positioning nuclear reactor internals in three-dimensional space according to claim 1, characterized in that: The model building module builds a three-dimensional space remote positioning device network model based on Yolov5 and Deeplabv3+ neural networks, and builds a training data set of the three-dimensional space remote positioning device network model based on the in-pile components; the training data set includes image data of the in-pile components at various distances and angles in the positioning site; The KNN module is used to fill in the missing depth information in the depth map; The super-resolution enlargement module is used to enrich the color image, increase the color vividness of the feature points, and make the depth information of the corresponding position in the depth image match the color image; The three-dimensional space information calculation module is used for three-dimensional calculation analysis and three-dimensional point cloud reconstruction.
5. A system for remotely positioning nuclear reactor internals in three-dimensional space according to claim 4, characterized in that: The model building module comprises: The first network construction unit, based on the target detection results, synchronously crops the color image and the depth image as the input image of the KNN module and the super-resolution enlargement module; The second network construction unit simultaneously crops the color image and the depth image based on the semantic segmentation result as the input image of the three-dimensional spatial information calculation module.
6. A system for remotely positioning nuclear reactor internals in three-dimensional space according to claim 1, characterized in that: The software system module includes a graphical user interface which is a software interface of a multi-phase array visual positioning system, and the interface is used for image display, configuration parameters, control and operation of the multi-phase array visual positioning system; wherein the home page includes a navigation bar at the top, function buttons below the navigation bar, a camera color image display area below the function buttons, a data display area below the camera color image display area, and a picture display area and a three-dimensional model display area on the right side of the home page.
7. A system for remotely positioning nuclear reactor internals in three-dimensional space according to claim 1, characterized in that: The system software interface also includes: Click the "Deformation Calculator" function button in the main view to enter the deformation calculator page. At the top of the deformation calculator page are the deformation calculation formula 1 description area, calculation input area and calculation result display area. The deformation calculation formula 1 is: Among them, ΔL represents the deformation of the object, F represents the force applied to the object, L represents the length of the object, Y represents the Young's modulus of the object, and A represents the cross-sectional area of the object; Below is the deformation calculation formula 2 description area, calculation input area and calculation result display area. The deformation calculation formula 2 is: Among them, ΔL represents the deformation of the object, σ represents the normal stress applied to the object, L represents the length of the object, and Y represents the Young's modulus of the object.
8. A system for remotely positioning nuclear reactor internals in three-dimensional space according to claim 1, characterized in that: The software system module has the functions of 3D analysis and calculation, 3D visualization, object deformation calculation, receiving and real-time display of image data from the camera, and also includes: Image display function: receiving an original image at a first preset time interval, the original image being taken by a binocular camera shooting unit in the nuclear reactor, and displaying the original image on the software; Positioning point distance measurement function: Generate a three-dimensional point cloud image from the color image and the depth image, identify the positioning points in the three-dimensional point cloud image, calculate the distance between two positioning points through the coordinates of the positioning points in the three-dimensional point cloud image, and display the positioning point distance in the software; 3D point cloud reconstruction image display function: read and display the 3D point cloud reconstruction image of the positioning bracket in the software at every second preset time interval; Target object trajectory point calculation and display function: calculate the center coordinates of the positioning bracket as the positioning bracket trajectory point, and save the trajectory point and display it in the software; 3D reconstruction function of the scene where the target object is located: obtain the color image and depth image of the scene where the positioning bracket is located, generate and save the 3D point cloud map of the scene through the color image and depth image; Shooting and timed shooting function: take color pictures and depth pictures through the binocular camera and save them; take color pictures and depth pictures and save them by setting the shooting interval through the software; Positioning point ranging function: control the binocular camera to capture color images and depth images and generate a 3D point cloud image, identify the positioning points, perform the positioning point ranging function and display the results in the software; Image display function: read the image in the disk and display it in the software; Format conversion function: read the raw format file, convert the raw format file into a png format file and save it; 3D point cloud image generation function: read the color image and depth image taken by the binocular camera, generate a 3D point cloud image through the captured color image and depth image, and save it; 3D point cloud point number statistics and free point selection distance measurement function: read the 3D point cloud, calculate the number of points in the 3D point cloud and display it; select any two points in the 3D point cloud by mouse to measure the distance and display it; Object deformation calculation function: by inputting the weight, length, Young's modulus and cross-sectional area of the object to be measured, the deformation of the object to be measured is calculated according to the given formula; by inputting the normal stress, length and Young's modulus of the object to be measured, the deformation of the object to be measured is calculated according to the given formula; Binocular camera depth map working mode selection function: select the depth map working mode of the binocular camera; There are three working modes: precision enhancement mode, long-distance enhancement mode, and frame rate enhancement mode.
Citation Information
Patent Citations
Target detection and visual positioning method based on YOLO series
CN112927297A
Part three-dimensional size measurement method and device based on point cloud data and medium
CN114998414A
Target ranging system based on self-supervised monocular depth estimation method
CN118172399A
Three-dimensional imaging method and apparatus, device, and storage medium
WO2024083006A1
