Depth camera external parameter calibration method and device and storage medium

Abstract

The invention discloses a depth camera external parameter calibration method and device and a storage medium, and relates to the technical field of camera calibration, and the method comprises the steps: obtaining a calibration box and scene depth point cloud data of a horizontal ground where the calibration box is located through a depth camera of a robot; extracting observation point cloud coordinates of a target vertex of the calibration box and an observation normal vector of the horizontal ground from the scene depth point cloud data; based on the observation point cloud coordinate, the observation normal vector, the reference coordinate of the target vertex and the reference normal vector of the horizontal ground, performing iterative optimization on the external parameter data of the depth camera to obtain the target external parameter of the depth camera, the reference coordinate of the target vertex being the coordinate of the target vertex calibrated in the robot coordinate system, and the reference normal vector of the horizontal ground being the coordinate of the target vertex. The reference normal vector is a normal vector calibrated by the horizontal ground in a robot coordinate system. The technical problem that the external parameter calibration effect is poor in the current mobile robot depth camera external parameter calibration technology is solved.

Description

Technical Field

[0001] This application relates to the field of camera calibration technology, and in particular to a method, apparatus and storage medium for calibrating extrinsic parameters of a depth camera. Background Technology

[0002] With the widespread application of mobile robot technology, the calibration of the precise coordinate relationship (i.e., extrinsic parameters) between the depth camera's installation position and the robot body is crucial, directly affecting the robot's localization, navigation, and operational accuracy. In current technological practices, common calibration methods rely solely on constraints from a single planar feature (such as the ground). While simple and fast, this provides limited constraint information and cannot solve for the complete spatial attitude and position parameters of the depth camera. Alternatively, multi-sensor data, such as LiDAR, can be introduced and registered with depth point clouds to increase constraints. While this improves calibration completeness, it complicates system configuration and calibration procedures, making it unsuitable for lightweight or low-cost robot platforms equipped only with a depth camera. Therefore, current mobile robot depth camera extrinsic parameter calibration technology suffers from poor calibration results.

[0003] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a method, device and storage medium for extrinsic parameter calibration of depth cameras, in order to solve the technical problem of poor extrinsic parameter calibration effect in the current extrinsic parameter calibration technology of depth cameras for mobile robots.

[0005] To achieve the above objectives, this application proposes a method for calibrating the extrinsic parameters of a depth camera, the method comprising: The robot's depth camera acquires scene depth point cloud data of the calibration box and the horizontal ground on which the calibration box is located. The observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground are extracted from the scene depth point cloud data. Based on the observed point cloud coordinates, the observed normal vector, the reference coordinates of the target vertex, and the reference normal vector of the horizontal ground, the extrinsic parameter data of the depth camera is iteratively optimized to obtain the target extrinsic parameters of the depth camera. The reference coordinates of the target vertex are the coordinates of the target vertex as calibrated in the robot coordinate system, and the reference normal vector is the normal vector of the horizontal ground as calibrated in the robot coordinate system.

[0006] In one embodiment, the step of extracting the observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data includes: A first point cloud portion and a second point cloud portion are separated from the scene depth point cloud data, wherein the first point cloud portion corresponds to the calibration box and the second point cloud portion corresponds to the horizontal ground. The observation point cloud coordinates of the target vertex of the calibration box are extracted based on the first point cloud portion; The observation normal vector of the horizontal ground is extracted based on the second point cloud portion.

[0007] In one embodiment, the step of extracting the observation point cloud coordinates of the target vertex of the calibration box based on the first point cloud portion includes: Planar feature analysis is performed on the first point cloud portion to obtain a first plane and a second plane. The first plane is the main surface of the calibration box facing the depth camera, and the second plane is the top surface of the calibration box connected to the main surface. The first plane and the second plane are perpendicular to each other. Obtain the spatial intersection line between the first plane and the second plane, and filter out a candidate point cloud set from the first point cloud portion, wherein the distance from each candidate point cloud in the candidate point cloud set to the spatial intersection line is less than a preset distance threshold; Calculate the spatial distance between any two candidate point clouds in the candidate point cloud set, determine the candidate point cloud pair with the largest spatial distance, and use the three-dimensional spatial coordinates of the candidate point cloud pair as the observation point cloud coordinates of the target vertex of the calibration box.

[0008] In one embodiment, the step of extracting the observation normal vector of the horizontal ground based on the second point cloud portion includes: The optimal fitting plane is calculated based on all data points in the second point cloud portion; The normal vector of the optimal fitting plane is extracted from the standard equation of the optimal fitting plane, and the normal vector is normalized to obtain the observation normal vector of the horizontal ground.

[0009] In one embodiment, the step of iteratively optimizing the extrinsic data of the depth camera based on the observed point cloud coordinates, the observed normal vector, the reference coordinates of the target vertex, and the reference normal vector of the horizontal ground to obtain the target extrinsic parameters of the depth camera includes: The observed point cloud coordinates are processed based on the extrinsic data of the depth camera to obtain transformed point cloud coordinates, and the positional deviation between the transformed point cloud coordinates and the reference coordinates is calculated. The observation normal vector is processed based on the extrinsic data of the depth camera to obtain the transformed normal vector, and the directional deviation between the transformed normal vector and the reference normal vector is calculated. The conversion loss is calculated based on the position deviation and the direction deviation, and the extrinsic parameter data is iteratively optimized based on the conversion loss to obtain the target extrinsic parameters.

[0010] In one embodiment, the step of processing the observed point cloud coordinates based on the extrinsic data of the depth camera to obtain transformed point cloud coordinates, and calculating the positional deviation between the transformed point cloud coordinates and the reference coordinates includes: The observed point cloud coordinates are rotated and translated using the external parameter data to obtain the transformed point cloud coordinates. The spatial straight-line distance between the transformed point cloud coordinates and the reference coordinates is calculated to obtain the position deviation.

[0011] In one embodiment, the step of processing the observed normal vector based on the extrinsic data of the depth camera to obtain a transformed normal vector, and calculating the directional deviation between the transformed normal vector and the reference normal vector includes: The observed normal vector is rotated using the extrinsic parameter data to obtain the transformed normal vector; The angular difference between the transformed normal vector and the reference normal vector is calculated to obtain the direction deviation.

[0012] In one embodiment, the step of calculating the conversion loss based on the position deviation and the orientation deviation, and iteratively optimizing the extrinsic parameter data based on the conversion loss to obtain the target extrinsic parameters includes: The conversion loss is obtained by accumulating the positional deviation and the orientation deviation; The extrinsic parameter data is adjusted based on the conversion loss to obtain new extrinsic parameter data; If the new extrinsic data satisfies the preset iterative optimization conditions, the new extrinsic data will be used as the target extrinsic data. If the new extrinsic data does not meet the preset iterative optimization conditions, the process of processing the observation point cloud coordinates based on the extrinsic data of the depth camera and subsequent steps are returned based on the new extrinsic data.

[0013] Furthermore, to achieve the above objectives, this application also proposes a depth camera extrinsic parameter calibration device, which includes: The data acquisition module is used to acquire scene depth point cloud data of the calibration box and the horizontal ground on which the calibration box is located through the robot's depth camera; The parameter extraction module is used to extract the observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data. The iterative optimization module is used to iteratively optimize the extrinsic parameter data of the depth camera based on the observed point cloud coordinates, the observed normal vector, the reference coordinates of the target vertex, and the reference normal vector of the horizontal ground to obtain the target extrinsic parameters of the depth camera. The reference coordinates of the target vertex are the coordinates of the target vertex as calibrated in the robot coordinate system, and the reference normal vector is the normal vector of the horizontal ground as calibrated in the robot coordinate system.

[0014] In addition, to achieve the above objectives, this application also proposes an electronic device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the depth camera extrinsic calibration method as described above.

[0015] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the depth camera extrinsic parameter calibration method described above.

[0016] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the depth camera extrinsic parameter calibration method described above.

[0017] This application provides a method for calibrating the extrinsic parameters of a depth camera. The method includes: acquiring scene depth point cloud data of a calibration box and the horizontal ground on which the calibration box is located using a robot's depth camera; extracting the observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data; and iteratively optimizing the extrinsic parameter data of the depth camera based on the observation point cloud coordinates, the observation normal vector, the reference coordinates of the target vertex, and the reference normal vector of the horizontal ground to obtain the target extrinsic parameters of the depth camera. The reference coordinates of the target vertex are the coordinates of the target vertex calibrated in the robot coordinate system, and the reference normal vector is the normal vector of the horizontal ground calibrated in the robot coordinate system.

[0018] This application acquires scene depth point cloud data including a calibration box and a horizontal ground using a depth camera, thereby obtaining multi-source 3D information including the structural features of the artificial calibration object (calibration box) and the features of the natural plane (horizontal ground). This solves the problem of insufficient constraint information from a single natural plane and provides the necessary multi-source data foundation for subsequent extrinsic parameter calculation. By extracting the observation point cloud coordinates of the target vertex on the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data, measurable spatial point constraints and planar orientation constraints are obtained from a single frame of data. This allows for the provision of reliable optimization targets for both the translation and rotation components of the extrinsic parameters. By comparing the extracted point cloud coordinates with preset reference coordinates to calculate the positional deviation, and by comparing the extracted ground normal vector information with preset normal vector information to calculate the orientational deviation, a dual metric system for evaluating both positional and orientational alignment is established. Based on these observations and combined with the reference coordinates and reference normal vector, the extrinsic parameter data from the depth camera is jointly optimized using an iterative optimization algorithm to solve for the target extrinsic parameters, ensuring the accurate solution of all degrees of freedom of the extrinsic parameter matrix. Compared to related solutions that mainly rely on the utilization of natural or specific geometric structures in the environment, this application introduces a simple calibration box as an aid, providing sufficient spatial point constraints without increasing system complexity, thereby enabling stable solution of extrinsic parameters. At the same time, it avoids the cumbersome process of complex data fusion and registration relying on additional sensors such as LiDAR. All data processing originates from the point cloud output of a single depth camera, realizing full-parameter calibration under pure visual perception conditions. While ensuring the integrity and accuracy of calibration, it significantly improves the simplicity and ease of operation of the calibration system. It is applicable to various mobile robot platforms equipped only with depth cameras, achieving fast, autonomous, and robust extrinsic parameter calibration, and improving the problem of poor extrinsic parameter calibration results in current technologies. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating an embodiment of the depth camera extrinsic parameter calibration method of this application. Figure 2 This is a flowchart illustrating Embodiment 2 of the depth camera extrinsic parameter calibration method of this application; Figure 3This is a flowchart illustrating Embodiment 3 of the depth camera extrinsic parameter calibration method of this application; Figure 4 This is a schematic diagram of the module structure of the depth camera extrinsic parameter calibration device according to an embodiment of this application; Figure 5 This is a schematic diagram of the device structure of the hardware operating environment involved in the depth camera extrinsic parameter calibration method in this application embodiment.

[0022] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0023] It should be understood that the first embodiment described herein is merely used to explain the technical solution of this application and is not intended to limit this application.

[0024] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0025] The main solution of the first embodiment of this application is as follows: acquiring scene depth point cloud data of the calibration box and the horizontal ground on which the calibration box is located through the robot's depth camera; extracting the observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data; iteratively optimizing the extrinsic parameter data of the depth camera based on the observation point cloud coordinates, the observation normal vector, the reference coordinates of the target vertex and the reference normal vector of the horizontal ground to obtain the target extrinsic parameters of the depth camera, wherein the reference coordinates of the target vertex are the coordinates of the target vertex calibrated in the robot coordinate system, and the reference normal vector is the normal vector of the horizontal ground calibrated in the robot coordinate system.

[0026] In the first embodiment, for ease of description, the following description uses the depth camera extrinsic parameter calibration device as the execution subject.

[0027] In existing technologies, common calibration methods mainly rely on the utilization of natural or specific geometric structures in the environment. For example, relying solely on a single planar feature (such as the ground) for constraints, while simple and fast, provides limited constraint information and cannot solve for the complete spatial pose and position parameters of the depth camera. Alternatively, introducing multi-sensor data such as LiDAR and registering it with the depth point cloud can increase constraints, which can improve calibration completeness, but leads to the complexity of the system configuration and calibration process, making it unsuitable for lightweight or low-cost robot platforms equipped only with a depth camera.

[0028] This application provides a solution that acquires scene depth point cloud data including a calibration box and a horizontal ground using a depth camera. This obtains multi-source 3D information, including the structural features of the artificial calibration object (calibration box) and the features of the natural plane (horizontal ground), solving the problem of insufficient constraint information from a single natural plane. This provides the necessary multi-source data foundation for subsequent extrinsic parameter calculation. By extracting the observation point cloud coordinates of the target vertex on the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data, measurable spatial point constraints and planar orientation constraints are obtained from a single frame of data. This allows for the provision of reliable optimization targets for both the translation and rotation components of the extrinsic parameters. By comparing the extracted point cloud coordinates with preset reference coordinates to calculate the positional deviation, and by comparing the extracted ground normal vector information with preset normal vector information to calculate the orientational deviation, a dual metric system for evaluating both positional and orientational alignment is established. Based on these observations and combined with the reference coordinates and reference normal vector, an iterative optimization algorithm is used to jointly optimize the extrinsic parameter data from the depth camera to solve for the target extrinsic parameters, ensuring the accurate solution of all degrees of freedom of the extrinsic parameter matrix. Compared to related solutions that mainly rely on the utilization of natural or specific geometric structures in the environment, this application introduces a simple calibration box as an aid, providing sufficient spatial point constraints without increasing system complexity, thereby enabling stable solution of extrinsic parameters. At the same time, it avoids the cumbersome process of complex data fusion and registration relying on additional sensors such as LiDAR. All data processing originates from the point cloud output of a single depth camera, realizing full-parameter calibration under pure visual perception conditions. While ensuring the integrity and accuracy of calibration, it significantly improves the simplicity and ease of operation of the calibration system. It is applicable to various mobile robot platforms equipped only with depth cameras, achieving fast, autonomous, and robust extrinsic parameter calibration, and improving the problem of poor extrinsic parameter calibration results in current technologies.

[0029] It should be noted that the executing entity in the first embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, mobile phone, or other electronic device, or a system, application, or program capable of implementing the above functions. The first embodiment and the following embodiments will be described using a depth camera extrinsic parameter calibration device as an example.

[0030] All actions involving the acquisition of signals, information, or data in this application are carried out in accordance with the relevant data protection laws and policies of the country where the application is located, and with the authorization of the owner of the relevant device.

[0031] Based on this, embodiments of this application provide a method for calibrating the extrinsic parameters of a depth camera, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the depth camera extrinsic parameter calibration method of this application.

[0032] In this embodiment, the depth camera extrinsic parameter calibration method includes steps S01~S03: Step S01: Obtain scene depth point cloud data of the calibration box and the horizontal ground on which the calibration box is located through the robot's depth camera.

[0033] It's important to note that a depth camera is a sensor capable of acquiring three-dimensional spatial information of a scene. It generates three-dimensional point cloud data with the camera itself as the origin by actively emitting light (such as infrared structured light) and receiving reflected signals. A calibration box is a physical calibration object with a known geometry, used to provide stable spatial structural features. The horizontal ground refers to the horizontal plane on which the robot is located. Scene depth point cloud data is a collection of three-dimensional spatial points acquired by the depth camera, including the calibration box and the horizontal ground. Each point has three-dimensional coordinates (x, y, z) in the depth camera's coordinate system.

[0034] Additionally, it should be noted that no other features (such as calibration plates, calibration points, etc.) need to be added to the calibration box, thus significantly reducing the difficulty of preparing the calibration material and facilitating the actual calibration work. Furthermore, the preferred geometric shape of the calibration box is a cuboid structure, such as a cuboid cardboard box. This is because other sharp cardboard boxes produce poor imaging quality in point clouds; for example, the vertices of a cone are difficult to detect in a point cloud.

[0035] Additionally, it should be noted that before starting external parameter calibration, the robot needs to be placed on a level surface, ensuring that there are no obstacles other than the calibration box within the robot's preset range (e.g., 5 meters), and the calibration box should be placed within the depth camera's field of view. Preferably, the depth camera should be placed diagonally above the calibration box, for example, approximately 1.5 meters in front of the calibration box and approximately 0.9 meters above it.

[0036] Understandably, step S01 acquires scene depth point cloud data of the calibration box and the horizontal ground through a depth camera, which solves the technical problem of insufficient constraint information caused by relying on only a single natural plane (such as the ground) in traditional methods, and provides a dual constraint basis including spatial points and planar directions for subsequent optimization.

[0037] Step S02: Extract the observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data.

[0038] It should be noted that the target vertices are pre-selected corner points on the calibration box, preferably the two vertices on the top surface of the calibration box directly facing the camera. The observed point cloud coordinates refer to the three-dimensional coordinates of the target vertices in the depth camera coordinate system. The observed normal vector is obtained by performing plane fitting on the point cloud data belonging to the horizontal ground plane in the scene depth point cloud data, resulting in the direction of the normal vector of that plane in the depth camera coordinate system.

[0039] Understandably, step S02 extracts the observation coordinates of the target vertex of the calibration box and the observation normal vector of the ground from the point cloud data, thereby simultaneously obtaining position and orientation constraints from a single frame of data. This provides reliable optimization targets for the translation and rotation matrices in the extrinsic parameters, avoiding the complexity of relying on multi-sensor data fusion such as lidar.

[0040] Step S03: Based on the observed point cloud coordinates, observed normal vector, reference coordinates of the target vertex, and reference normal vector of the horizontal ground, iteratively optimize the extrinsic data of the depth camera to obtain the target extrinsic parameters of the depth camera. The reference coordinates of the target vertex are the coordinates of the target vertex as calibrated in the robot coordinate system, and the reference normal vector is the normal vector of the horizontal ground as calibrated in the robot coordinate system.

[0041] It should be noted that the robot coordinate system is a reference coordinate system fixed on the robot body. The origin can be the robot's geometric center or a pre-selected mounting point, used to describe the spatial relationship between the robot and objects in the environment. Reference coordinates refer to the known 3D coordinates of the target vertex in the robot coordinate system, measured or calibrated beforehand. The reference normal vector refers to the known direction of the normal vector of the horizontal ground in the robot coordinate system, preferably vertically upward, and is represented as N = (1, 1, 1). Extrinsic parameter data is a set of parameters describing the pose relationship between the depth camera coordinate system and the robot coordinate system, including the rotation matrix R and the translation matrix. The target extrinsic parameters are the optimal extrinsic parameter data obtained through iterative optimization, ensuring that the observed point cloud coordinates are aligned with the reference coordinates after transformation to the robot coordinate system, and that the observed normal vector is aligned with the reference normal vector after transformation.

[0042] Understandably, step S03 iteratively optimizes the extrinsic parameters based on the observed and reference values, achieving an accurate solution for the extrinsic parameter matrix. Without introducing additional sensors, calibration can be completed using only depth camera point clouds and ordinary calibration boxes, ensuring the integrity of the calibration while significantly simplifying the system structure and operation process. It is applicable to various mobile robot platforms equipped only with depth cameras, effectively improving the accuracy, robustness, and practicality of extrinsic parameter calibration.

[0043] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 In step S12, the steps of extracting the observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data include steps S11 to S13: Step S11: Separate the first point cloud portion and the second point cloud portion from the scene depth point cloud data. The first point cloud portion corresponds to the calibration box, and the second point cloud portion corresponds to the horizontal ground.

[0044] It should be noted that the first point cloud component is a set of all 3D data points belonging to the surface of the calibration box, separated from the scene depth point cloud data through spatial segmentation, clustering, or identification methods based on prior geometric models. The second point cloud component is a set of all 3D data points belonging to the horizontal ground, separated from the scene depth point cloud data.

[0045] Step S12: Extract the observation point cloud coordinates of the target vertex of the calibration box based on the first point cloud portion.

[0046] In one feasible implementation, step S12 includes steps A01 to A03: Step A01: Perform planar feature analysis on the first point cloud portion to obtain a first plane and a second plane. The first plane is the main surface of the calibration box facing the depth camera, and the second plane is the top surface of the calibration box connected to the main surface. The first plane and the second plane are perpendicular to each other.

[0047] It should be noted that planar feature analysis is a process of mathematically processing the first point cloud portion to identify and extract planar structures within it. This can be achieved through iterative fitting algorithms (such as the random sampling consensus algorithm) or region-growing-based segmentation algorithms, aiming to reconstruct geometric planes from discrete point cloud data. The first plane is the plane identified from the first point cloud portion through planar feature analysis, corresponding to the main surface of the calibration box facing the depth camera. The second plane is another plane identified from the first point cloud portion through planar feature analysis, connected to the first plane, corresponding to the top surface of the calibration box connected to the main surface.

[0048] Step A02: Obtain the spatial intersection line between the first plane and the second plane, and filter out a candidate point cloud set from the first point cloud portion, wherein the distance from each candidate point cloud in the candidate point cloud set to the spatial intersection line is less than a preset distance threshold.

[0049] It should be noted that the spatial intersection line is a three-dimensional straight line obtained by solving the simultaneous mathematical equations of the first and second planes. This straight line is the actual boundary line between the first and second planes in three-dimensional space. The candidate point cloud set is a subset selected from the first point cloud portion. Each candidate point cloud in this subset satisfies the geometric condition that its perpendicular distance to the spatial intersection line is less than a preset distance threshold. A candidate point cloud is a single three-dimensional data point in the candidate point cloud set. The preset distance threshold is a pre-set positive numerical parameter used to control the leniency of the selection of candidate point clouds, for example, 0.02 meters.

[0050] Step A03: Calculate the spatial distance between any two candidate point clouds in the candidate point cloud set, determine the candidate point cloud pair with the largest spatial distance, and use the three-dimensional spatial coordinates of the candidate point cloud pair as the observation point cloud coordinates of the target vertex of the calibration box.

[0051] It should be noted that spatial distance refers to the straight-line distance between two points (candidate point clouds) in three-dimensional Euclidean space. A candidate point cloud pair is a combination of any two candidate point clouds selected from the candidate point cloud set. Three-dimensional spatial coordinates refer to a set of values ​​representing the spatial position of a candidate point cloud in the depth camera coordinate system, which can be represented as (x, y, z).

[0052] Additionally, it should be noted that since the depth camera relies on emitted light to detect point cloud data, it can only acquire 3D data facing the sensor surface. The depth camera needs to be placed diagonally above the calibration box. Therefore, only the main surface and top surface facing the depth camera in the calibration box can form an effective point cloud. Furthermore, the bottom surface of the calibration box is in contact with the horizontal ground, so its point cloud blends with the ground, making it difficult to distinguish and blurry. Therefore, it is not used as a feature source. The two vertices on the top surface of the calibration box facing the camera are chosen because they are clearly imaged in the point cloud, have stable positions, and the line segment formed by the two points can provide clear spatial direction and distance constraints, thus providing sufficient and necessary observation information for extrinsic parameter optimization.

[0053] For example, planar feature analysis is performed on the first point cloud portion to identify two mutually perpendicular surfaces: the main surface facing the depth camera (first plane) and the top surface connected to it (second plane). The spatial intersection of the two planes is calculated, and all points in the first point cloud portion whose distance to this intersection is less than a preset distance threshold are selected to obtain a candidate point cloud set. Finally, the Euclidean distance between any two candidate point clouds in the candidate point cloud set is calculated, and the pair of points with the largest distance is selected, and their three-dimensional coordinates are determined as the observation point cloud coordinates of the two target vertices of the calibration box. This process, through multi-plane geometric constraints and the maximum distance criterion, achieves robust and accurate extraction of the box feature vertices even when the point cloud contains noise or partial missing data, providing key and reliable spatial point constraints for subsequent extrinsic parameter optimization.

[0054] Step S13: Extract the observation normal vector of the horizontal ground based on the second point cloud portion.

[0055] In one feasible implementation, step S13 includes steps A11-A12: Step A11: Calculate the optimal fitting plane based on all data points in the second point cloud portion.

[0056] It should be noted that the optimal fitting plane is an infinitely extending two-dimensional plane that best represents the spatial distribution trend of all data points in the least squares sense. This plane minimizes the sum of the squares of the vertical distances from all data points to it, and can be determined by Principal Component Analysis (PCA) or least squares plane fitting algorithms.

[0057] Step A12: Extract the normal vector of the best-fit plane from the standard equation of the best-fit plane, and normalize the normal vector to obtain the observation normal vector of the horizontal ground.

[0058] It should be noted that the standard equation is the mathematical expression for a plane in three-dimensional space, in the form Ax + By + Cz + D = 0. The vector (A, B, C) formed by the coefficients is the normal vector of the best-fit plane, and this vector is perpendicular to the best-fit plane. Normalization is a mathematical operation that converts non-zero vectors into unit vectors (with a magnitude of 1) in the same direction.

[0059] For example, using least squares fitting algorithms, the optimal fitting plane is calculated based on all data points in the second point cloud to eliminate local undulations and noise interference. The normal vector (A, B, C) is extracted from the standard equation Ax + By + Cz + D = 0 of the optimal fitting plane and normalized to obtain the unit normal vector as the observation normal vector. This process provides stable and accurate direction observations to address the problem of insufficient single-plane constraint information, laying a reliable foundation for optimizing the extrinsic rotation parameters.

[0060] In this embodiment, the scene depth point cloud data is separated into a first point cloud portion (calibration box portion) and a second point cloud portion (horizontal ground portion) to solve the problem of mutual interference between mixed data and provide clean input for feature extraction. Candidate point cloud pairs are extracted from the calibration box point cloud through planar feature analysis, intersection constraint screening, and the maximum distance criterion, thereby obtaining the observation point cloud coordinates of the target vertex. This overcomes the technical deficiency that a single plane cannot provide reliable spatial point constraints, laying the foundation for extrinsic parameter optimization. By performing overall planar fitting and normalization on the second point cloud portion, stable and accurate observation normal vectors are obtained, solving the problem of local noise interference and providing reliable directional constraints for extrinsic parameter optimization. The above steps work together, utilizing only a single depth camera data, while providing precise point and surface dual geometric constraints, thus enabling complete and high-precision solution of extrinsic parameters. This effectively overcomes the shortcomings of insufficient constraints or overly complex systems in existing technologies.

[0061] Based on the first and second embodiments of this application, in the third embodiment of this application, the content that is the same as or similar to the first and second embodiments described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 In step 03, the steps of iteratively optimizing the extrinsic data of the depth camera based on the observed point cloud coordinates, observed normal vectors, reference coordinates of the target vertex, and reference normal vector of the horizontal ground to obtain the target extrinsic parameters of the depth camera include steps S21 to S23: Step S21: Process the observed point cloud coordinates based on the extrinsic data of the depth camera to obtain the transformed point cloud coordinates, and calculate the positional deviation between the transformed point cloud coordinates and the reference coordinates.

[0062] It should be noted that transformed point cloud coordinates are the three-dimensional coordinates obtained by transforming the observed point cloud coordinates from the camera coordinate system to the robot coordinate system. Position deviation refers to the spatial straight-line distance between the transformed point cloud coordinates and the reference coordinates, which can be calculated using Euclidean distance.

[0063] In one feasible implementation, step S11 includes steps B01~B02: Step B01: Rotate and translate the observed point cloud coordinates using external parameter data to obtain the transformed point cloud coordinates.

[0064] Step B02: Calculate the spatial straight-line distance between the transformed point cloud coordinates and the reference coordinates to obtain the positional deviation.

[0065] For example, the formula for calculating the positional deviation is:

[0066] in, For positional deviation, The rotation matrix is ​​in the extrinsic parameter data. To observe the coordinates of the point cloud, This indicates that the coordinates of the observed point cloud have been rotated. The translation matrix in the extrinsic data. This indicates that the coordinates of the observed point cloud have been translated, and the result is the corresponding representation of the transformed point cloud coordinates. Used as reference coordinates.

[0067] Step S22: Process the observed normal vector based on the extrinsic data of the depth camera to obtain the transformed normal vector, and calculate the directional deviation between the transformed normal vector and the reference normal vector.

[0068] It should be noted that the transformed normal vector is a unit normal vector obtained by rotating and transforming the observed normal vector, used to align the observed ground orientation to the reference coordinate system. Orientation deviation refers to the angular difference between the transformed normal vector and the reference normal vector, which can be obtained by the cosine of the angle between the vectors or by directly calculating the angle in radians.

[0069] In one feasible implementation, step S12 includes steps B11-B12: Step B11: Rotate the observed normal vector using extrinsic data to obtain the transformed normal vector.

[0070] Step B12: Calculate the angular difference between the transformed normal vector and the reference normal vector to obtain the direction deviation.

[0071] For example, the formula for calculating directional deviation is:

[0072] in, For directional deviation, The rotation matrix is ​​in the extrinsic parameter data. To observe the normal vector, This indicates that the observation normal vector is rotated. This is the reference normal vector.

[0073] Step S23: Calculate the conversion loss based on the position deviation and direction deviation, and iteratively optimize the extrinsic parameter data based on the conversion loss to obtain the target extrinsic parameters.

[0074] It should be noted that the transformation loss is a scalar value obtained by aggregating positional and directional deviations according to a predetermined rule (such as addition). It is used to evaluate the overall error of the current extrinsic data in terms of point and surface alignment, and to guide iterative optimization algorithms (such as gradient descent) to adjust the extrinsic data.

[0075] In one feasible implementation, step S13 includes steps B21-B24: Step B21: Accumulate the positional and directional deviations to obtain the conversion loss.

[0076] For example, the conversion loss can be expressed as:

[0077] in, For conversion loss, The extrinsic parameters include rotation and translation matrices. For positional deviation, This is a directional deviation.

[0078] Step B22: Adjust the extrinsic parameter data based on the conversion loss to obtain new extrinsic parameter data.

[0079] Step B23: If the new extrinsic data meets the preset iterative optimization conditions, the new extrinsic data is used as the target extrinsic data.

[0080] It should be noted that the preset iterative optimization conditions are a set of pre-defined numerical or logical criteria used to determine whether an acceptable solution has been reached during the iterative optimization process, thus terminating the iteration to avoid infinite loops or invalid calculations. These criteria can include the reduction in transformation loss being less than a preset threshold (e.g., ...). ), the number of iterations to obtain the maximum number of iterations (e.g., 100 times), etc.

[0081] Step B24: If the new extrinsic data does not meet the preset iterative optimization conditions, return to the step of processing the observation point cloud coordinates based on the extrinsic data of the depth camera, and follow-up steps.

[0082] For example, after obtaining the transformation loss, the Jacobian matrix of the transformation loss with respect to the extrinsic data is calculated:

[0083] in, For conversion loss, For the extrinsic parameter data to be optimized, Transformation loss Optimizing external parameter data The Jacobian matrix is ​​a matrix representation of the derivative of a multivariate function, used to describe the local gradient information of the transformation loss as the extrinsic parameters change.

[0084] After obtaining the Jacobian matrix, we further construct the normal equation (the normal equation is an analytical method for directly solving for the optimal parameters in linear regression problems; by minimizing the condition that the gradient of the loss function is zero, the closed-form solution of the parameters can be derived):

[0085] in, Jacobian matrix transpose, The information matrix, which approximates the Hessian matrix, is an approximation of the second derivative of the transformation loss. Let be the gradient vector, representing the direction of descent of the transformation loss at the current extrinsic data. This refers to the update amount of the extrinsic parameter data to be optimized, i.e., in which direction and by how much the extrinsic parameters should be adjusted in this iteration. Based on... Adjust the external parameter data to obtain new external parameter data, and determine whether the external parameter data meets the preset iterative optimization conditions (the decrease in transformation loss is less than the preset threshold). If it meets the conditions, output it as the target external parameter; if it does not meet the conditions, use the new external parameter data as the target external parameter, return to the steps of processing the observation point cloud coordinates based on the external parameter data of the depth camera and subsequent steps, and continue iterating.

[0086] In this embodiment, the spatial distance error between the camera observation point and the reference point is quantified by calculating the position deviation, which directly constrains the translation component of the extrinsic parameters. The angular error between the observed ground normal vector and the reference normal vector is quantified by calculating the direction deviation, which directly constrains the rotation component of the extrinsic parameters. The transformation loss is constructed by combining the two types of deviations, and the extrinsic parameters are jointly updated by an iterative optimization algorithm until the convergence condition is met. This process achieves high-precision solution of extrinsic parameters by using only a monocular depth camera under the condition of point and surface dual constraints, overcoming the technical limitations of insufficient constraint information in a single plane and the complexity of multi-sensor systems.

[0087] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the depth camera extrinsic parameter calibration method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0088] This application also provides a depth camera extrinsic parameter calibration device; please refer to... Figure 4 The depth camera extrinsic calibration device includes: Data acquisition module 10 is used to acquire scene depth point cloud data of the calibration box and the horizontal ground where the calibration box is located through the robot's depth camera; The parameter extraction module 20 is used to extract the observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data. The iterative optimization module 30 is used to iteratively optimize the extrinsic parameter data of the depth camera based on the observation point cloud coordinates, the observation normal vector, the reference coordinates of the target vertex, and the reference normal vector of the horizontal ground to obtain the target extrinsic parameters of the depth camera. The reference coordinates of the target vertex are the coordinates of the target vertex as calibrated in the robot coordinate system, and the reference normal vector is the normal vector of the horizontal ground as calibrated in the robot coordinate system.

[0089] The depth camera extrinsic parameter calibration device provided in this application, employing the depth camera extrinsic parameter calibration method described in the above embodiments, can solve the technical problem of poor extrinsic parameter calibration results in current mobile robot depth camera extrinsic parameter calibration technology. Compared with the prior art, the beneficial effects of the depth camera extrinsic parameter calibration device provided in this application are the same as those of the depth camera extrinsic parameter calibration method provided in the above embodiments, and other technical features in the depth camera extrinsic parameter calibration device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0090] This application provides an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the depth camera extrinsic calibration method in Embodiment 1 above.

[0091] The following is for reference. Figure 5 The diagram illustrates a structural schematic of an electronic device suitable for implementing embodiments of this application. The electronic devices in these embodiments may include, but are not limited to, mobile terminals such as robots, unmanned vehicles, and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as desktop computers. Figure 5 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0092] like Figure 5 As shown, the electronic device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the electronic device. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. The communication device 1009 allows the electronic device to communicate wirelessly or wiredly with other devices to exchange data. Although the diagrams show electronic devices with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented alternatively.

[0093] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0094] The electronic device provided in this application, employing the depth camera extrinsic parameter calibration method described in the above embodiments, can solve the technical problem of poor extrinsic parameter calibration results in current mobile robot depth camera extrinsic parameter calibration technology. Compared with the prior art, the beneficial effects of the electronic device provided in this application are the same as those of the depth camera extrinsic parameter calibration method provided in the above embodiments, and other technical features of this electronic device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0095] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0096] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0097] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the depth camera extrinsic calibration method in the above embodiments.

[0098] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0099] The aforementioned computer-readable storage medium may be included in an electronic device or may exist independently without being assembled into an electronic device.

[0100] The aforementioned computer-readable storage medium carries one or more programs that, when executed by an electronic device, cause the depth camera extrinsic calibration device to: acquire scene depth point cloud data of the calibration box and the horizontal ground on which the calibration box is located through the robot's depth camera; extract the observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data; and iteratively optimize the extrinsic parameter data of the depth camera based on the observation point cloud coordinates, the observation normal vector, the reference coordinates of the target vertex, and the reference normal vector of the horizontal ground to obtain the target extrinsic parameters of the depth camera, wherein the reference coordinates of the target vertex are the coordinates of the target vertex calibrated in the robot coordinate system, and the reference normal vector is the normal vector of the horizontal ground calibrated in the robot coordinate system.

[0101] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0102] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and flowcharts, and combinations of blocks in the block diagrams and flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0103] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0104] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned depth camera extrinsic parameter calibration method. This solves the technical problem of poor extrinsic parameter calibration results in current mobile robot depth camera extrinsic parameter calibration technologies. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the depth camera extrinsic parameter calibration method provided in the above embodiments, and will not be elaborated upon here.

[0105] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the depth camera extrinsic calibration method as described above.

[0106] The computer program product provided in this application can solve the technical problem of poor extrinsic parameter calibration effect in the current extrinsic parameter calibration technology for depth cameras in mobile robots. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the depth camera extrinsic parameter calibration method provided in the above embodiments, and will not be repeated here.

[0107] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for calibrating extrinsic parameters of a depth camera, characterized in that, The depth camera extrinsic parameter calibration method includes: The robot's depth camera acquires scene depth point cloud data of the calibration box and the horizontal ground on which the calibration box is located. The observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground are extracted from the scene depth point cloud data. Based on the observed point cloud coordinates, the observed normal vector, the reference coordinates of the target vertex, and the reference normal vector of the horizontal ground, the extrinsic parameter data of the depth camera is iteratively optimized to obtain the target extrinsic parameters of the depth camera. The reference coordinates of the target vertex are the coordinates of the target vertex as calibrated in the robot coordinate system, and the reference normal vector is the normal vector of the horizontal ground as calibrated in the robot coordinate system.

2. The depth camera extrinsic parameter calibration method as described in claim 1, characterized in that, The steps of extracting the observation point cloud coordinates of the target vertex of the calibration box and the observation normal vector of the horizontal ground from the scene depth point cloud data include: A first point cloud portion and a second point cloud portion are separated from the scene depth point cloud data, wherein the first point cloud portion corresponds to the calibration box and the second point cloud portion corresponds to the horizontal ground. The observation point cloud coordinates of the target vertex of the calibration box are extracted based on the first point cloud portion; The observation normal vector of the horizontal ground is extracted based on the second point cloud portion.

3. The depth camera extrinsic parameter calibration method as described in claim 2, characterized in that, The step of extracting the observation point cloud coordinates of the target vertex of the calibration box based on the first point cloud portion includes: Planar feature analysis is performed on the first point cloud portion to obtain a first plane and a second plane. The first plane is the main surface of the calibration box facing the depth camera, and the second plane is the top surface of the calibration box connected to the main surface. The first plane and the second plane are perpendicular to each other. Obtain the spatial intersection line between the first plane and the second plane, and filter out a candidate point cloud set from the first point cloud portion, wherein the distance from each candidate point cloud in the candidate point cloud set to the spatial intersection line is less than a preset distance threshold; Calculate the spatial distance between any two candidate point clouds in the candidate point cloud set, determine the candidate point cloud pair with the largest spatial distance, and use the three-dimensional spatial coordinates of the candidate point cloud pair as the observation point cloud coordinates of the target vertex of the calibration box.

4. The depth camera extrinsic parameter calibration method as described in claim 2, characterized in that, The step of extracting the observation normal vector of the horizontal ground based on the second point cloud portion includes: The optimal fitting plane is calculated based on all data points in the second point cloud portion; The normal vector of the optimal fitting plane is extracted from the standard equation of the optimal fitting plane, and the normal vector is normalized to obtain the observation normal vector of the horizontal ground.

5. The depth camera extrinsic parameter calibration method as described in claim 1, characterized in that, The step of iteratively optimizing the extrinsic data of the depth camera based on the observed point cloud coordinates, the observed normal vector, the reference coordinates of the target vertex, and the reference normal vector of the horizontal ground to obtain the target extrinsic parameters of the depth camera includes: The observed point cloud coordinates are processed based on the extrinsic data of the depth camera to obtain transformed point cloud coordinates, and the positional deviation between the transformed point cloud coordinates and the reference coordinates is calculated. The observation normal vector is processed based on the extrinsic data of the depth camera to obtain the transformed normal vector, and the directional deviation between the transformed normal vector and the reference normal vector is calculated. The conversion loss is calculated based on the position deviation and the direction deviation, and the extrinsic parameter data is iteratively optimized based on the conversion loss to obtain the target extrinsic parameters.

6. The depth camera extrinsic parameter calibration method as described in claim 5, characterized in that, The step of processing the observed point cloud coordinates based on the extrinsic data of the depth camera to obtain transformed point cloud coordinates, and calculating the positional deviation between the transformed point cloud coordinates and the reference coordinates includes: The observed point cloud coordinates are rotated and translated using the external parameter data to obtain the transformed point cloud coordinates. The spatial straight-line distance between the transformed point cloud coordinates and the reference coordinates is calculated to obtain the position deviation.

7. The depth camera extrinsic parameter calibration method as described in claim 5, characterized in that, The step of processing the observed normal vector based on the extrinsic data of the depth camera to obtain the transformed normal vector, and calculating the direction deviation between the transformed normal vector and the reference normal vector includes: The observed normal vector is rotated using the extrinsic parameter data to obtain the transformed normal vector; The angular difference between the transformed normal vector and the reference normal vector is calculated to obtain the direction deviation.

8. The depth camera extrinsic parameter calibration method as described in claim 5, characterized in that, The steps of calculating the conversion loss based on the position deviation and the direction deviation, and iteratively optimizing the extrinsic parameter data based on the conversion loss to obtain the target extrinsic parameters include: The conversion loss is obtained by accumulating the positional deviation and the orientation deviation; The extrinsic parameter data is adjusted based on the conversion loss to obtain new extrinsic parameter data; If the new extrinsic data satisfies the preset iterative optimization conditions, the new extrinsic data will be used as the target extrinsic data. If the new extrinsic data does not meet the preset iterative optimization conditions, the process of processing the observation point cloud coordinates based on the extrinsic data of the depth camera and subsequent steps are returned based on the new extrinsic data.

9. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored on the memory and executable on the processor, the computer program being configured to implement the steps of the depth camera extrinsic calibration method as described in any one of claims 1 to 8.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the depth camera extrinsic parameter calibration method as described in any one of claims 1 to 8.

Images