Camera external parameter calibration method and device, vehicle and computer readable storage medium

Abstract

The invention discloses a camera external parameter calibration method and device, a vehicle and a computer readable storage medium, and the method comprises the steps: obtaining in-vehicle image data collected by a camera in the vehicle, obtaining a two-dimensional coordinate point set Q corresponding to a target key pixel point in the vehicle based on the image data, carrying out the error correction of the two-dimensional coordinate point set Q, and obtaining a target key pixel point in the vehicle; and obtaining a two-dimensional coordinate point set S after error correction. And then acquiring a three-dimensional coordinate point set P corresponding to the corrected two-dimensional coordinate point set S, and calibrating external parameters of the camera according to the corrected two-dimensional coordinate point set S and the three-dimensional coordinate point set P. According to the method, the external parameters of the camera after the position of the camera is changed can be calculated more accurately, and the accuracy of follow-up applications such as AR-HUD and sight line estimation is greatly improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a camera extrinsic calibration method, device, vehicle and computer-readable storage medium. Background Art

[0002] For current on-board cameras, the accuracy of camera extrinsics is crucial to the user experience, so on-board cameras need to be calibrated frequently. When the camera position changes, the captured image will also change accordingly, resulting in changes in the two-dimensional coordinate positions of key points in the image, which indirectly changes the correspondence between the two-dimensional coordinate points and the three-dimensional coordinate points, thus affecting the accuracy of the camera extrinsic calibration. However, the existing calibration method is inefficient and affects the user experience. Summary of the invention

[0003] In view of this, embodiments of the present invention provide a camera extrinsic parameter calibration method, device, vehicle and computer-readable storage medium.

[0004] Specifically, the embodiments of the present disclosure are implemented through the following technical solutions:

[0005] According to a first aspect of the present invention, a camera extrinsic calibration method is provided, the method comprising:

[0006] Acquire image data inside the vehicle collected by a camera inside the vehicle;

[0007] Based on the image data, a two-dimensional coordinate point set Q corresponding to a target key pixel point in the vehicle is obtained, and error correction processing is performed on the two-dimensional coordinate point set Q to obtain a two-dimensional coordinate point set S after error correction;

[0008] A three-dimensional coordinate point set P corresponding to the error-corrected two-dimensional coordinate point set S is obtained, and the external parameters of the camera are calibrated according to the error-corrected two-dimensional coordinate point set S and the three-dimensional coordinate point set P.

[0009] Optionally, error correction processing is performed on the two-dimensional coordinate point set Q to obtain a two-dimensional coordinate point set S after error correction, including:

[0010] Determine the initial image area where the target key pixel point is located according to the two-dimensional coordinate point set Q;

[0011] The initial image region is subjected to image error correction processing to obtain a target image region, and the error-corrected two-dimensional coordinate point set S is obtained according to the target image region and the two-dimensional coordinate point set Q.

[0012] Optionally, obtaining a two-dimensional coordinate point set Q of target key pixel points in the vehicle based on the image data includes:

[0013] Obtaining target key pixel points of the target area in the vehicle based on the image data and a preset model prediction;

[0014] The two-dimensional coordinate point set Q is generated according to the two-dimensional coordinates of the target key pixel points.

[0015] Optionally, determining an initial image region where the target key pixel point is located according to the two-dimensional coordinate point set Q;

[0016] The initial image region is subjected to image error correction processing to obtain a target image region, and a corresponding two-dimensional coordinate point set S is generated according to the target image region.

[0017] Optionally, target key pixel points of the target area in the vehicle are predicted based on the image data and a preset model.

[0018] Optionally, calibrating the extrinsic parameters of the camera according to the error-corrected two-dimensional coordinate point set S and the three-dimensional coordinate point set P includes:

[0019] Repeat the following steps until the preset conditions are met:

[0020] Performing curve fitting on the error-corrected two-dimensional coordinate point set S to obtain at least one curve;

[0021] For the at least one curve, a point sampling operation is performed according to a preset rule to obtain a new two-dimensional coordinate point set H; wherein the preset rule is an interval sampling method or a random sampling method;

[0022] The three-dimensional coordinate point set P is transformed by the camera external parameters calibrated last time to obtain a two-dimensional coordinate point set B, and the point closest to each coordinate point in the two-dimensional coordinate point set B is found in the new two-dimensional coordinate point set H, and a two-dimensional coordinate point set G is generated;

[0023] Using the two-dimensional coordinate point set G and the three-dimensional coordinate point set P, updating the external parameters of the camera;

[0024] Among them, the external parameters obtained when the preset conditions are met are the final external parameters determined.

[0025] Optionally, the preset condition is that a change in an external parameter obtained by a current iterative process compared to an external parameter obtained by a previous iterative process is less than a preset threshold; or, the preset condition is that the number of iterations exceeds a preset threshold.

[0026] Optionally, obtaining a three-dimensional coordinate point set P corresponding to the error-corrected two-dimensional coordinate point set S includes:

[0027] According to the vehicle model data of the vehicle, determine the vehicle component where the key point corresponding to the target key pixel point is located; wherein the vehicle model data includes the three-dimensional coordinate position information of the point of the vehicle component of the vehicle;

[0028] Determine the three-dimensional coordinate position information of the key point according to the vehicle component; wherein the three-dimensional coordinate position information indicates the three-dimensional coordinates of the key point;

[0029] The three-dimensional coordinate point set P is generated according to each of the three-dimensional coordinate position information.

[0030] Optionally, the image data is sent to a cloud server; wherein the image data is used to perform a calibration operation of the camera's external parameters.

[0031] According to a second aspect of the present invention, a camera extrinsic calibration device is provided, comprising:

[0032] An acquisition module, used for acquiring image data inside the vehicle captured by a camera inside the vehicle;

[0033] An image processing module, used for obtaining a two-dimensional coordinate point set Q corresponding to the target key pixel points in the vehicle based on the image data, and performing error correction processing on the two-dimensional coordinate point set Q to obtain a two-dimensional coordinate point set S after error correction;

[0034] A mapping module is used to obtain a three-dimensional coordinate point set P corresponding to the error-corrected two-dimensional coordinate point set S, and calibrate the external parameters of the camera according to the error-corrected two-dimensional coordinate point set S and the three-dimensional coordinate point set P.

[0035] According to a third aspect of the present invention, there is provided a vehicle, the vehicle comprising: a camera, at least one processor, and a memory communicatively connected to the at least one processor; wherein the camera is used to collect image data in the vehicle;

[0036] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method as described in any implementation of the first method.

[0037] According to a fourth aspect of the present invention, there is provided a computer-readable storage medium having computer instructions, which, when executed, implement the method described in any embodiment of the first method.

[0038] The camera extrinsic parameter calibration provided in the technical solution of the embodiment of the present disclosure obtains the image data inside the vehicle collected by the camera inside the vehicle, obtains the two-dimensional coordinate point set Q corresponding to the key pixel points of the target inside the vehicle based on the image data, and performs error correction processing on the two-dimensional coordinate point set Q to obtain the two-dimensional coordinate point set S after error correction. Then, the three-dimensional coordinate point set P corresponding to the two-dimensional coordinate point set S after error correction is obtained, and the camera extrinsic parameters are calibrated according to the two-dimensional coordinate point set S and the three-dimensional coordinate point set P after error correction. The camera extrinsic parameters can be calculated more accurately after the camera position changes, which greatly improves the accuracy of subsequent applications such as AR-HUD and line of sight estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in one or more embodiments of the present disclosure or related technologies, the drawings required for use in the embodiments or related technology descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in one or more embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0040] Figure 1 is a flow chart of a camera extrinsic calibration method shown in an embodiment of the present application;

[0041] Figure 2 is a schematic diagram of an image before error correction processing shown in an embodiment of the present application;

[0042] Figure 3 is a schematic diagram of an image after error correction processing shown in an embodiment of the present application;

[0043] Figure 4 is a structural block diagram of a camera extrinsic calibration device shown in an embodiment of the present application;

[0044] Figure 5 It is a structural block diagram of a vehicle shown in an embodiment of the present application. DETAILED DESCRIPTION

[0045] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with this specification. Instead, they are merely examples of devices and methods consistent with some aspects of this specification as detailed in the appended claims.

[0046] The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit this specification. The singular forms "a", "the" and "the" used in this specification and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0047] It should be understood that although the terms first, second, third, etc. may be used in this specification to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this specification, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0048] The camera extrinsic parameters change according to the position of the camera. To accurately estimate the camera extrinsic parameters, the camera dynamic calibration algorithm is required. As the camera position changes, the captured image also changes accordingly, which causes the two-dimensional coordinate position of the key pixel points in the image to change, indirectly causing the correspondence between the two-dimensional coordinate points and the three-dimensional coordinate points to change, thereby calculating different camera extrinsic parameters. The accuracy of camera extrinsic parameters is an important part of other related applications, and will seriously affect the accuracy of subsequent related applications, such as AR-HUD adjustment, line of sight estimation in driver distraction, etc., and ultimately seriously affect the user experience.

[0049] like Figure 1 As shown, Figure 1 : is a flow chart of a camera extrinsic parameter calibration method shown in an embodiment of the present application. The method can be used to accurately calculate the camera extrinsic parameters after the camera position changes, and includes the following steps:

[0050] In step 100 , image data inside the vehicle is acquired by a camera inside the vehicle.

[0051] In this embodiment, the in-vehicle camera can be placed on the A-pillar, rearview mirror, or steering wheel in the vehicle. The camera can be an IR camera or an RGB camera with a resolution of more than 1 million. The in-vehicle camera captures the video stream in the vehicle cabin in real time for input into the algorithm in the subsequent steps.

[0052] In step 101, a two-dimensional coordinate point set Q corresponding to target key pixel points in the vehicle is obtained based on image data, and error correction processing is performed on the two-dimensional coordinate point set Q to obtain a two-dimensional coordinate point set S after error correction.

[0053] In the present embodiment, a prediction model for target key pixel points in a vehicle cabin is provided, and target key pixel points of a target area in a vehicle can be obtained based on image data and the prediction model. The key pixel point prediction model uses a dense prediction network to simultaneously predict a group of key point objects and a group of posture objects. The key pixel point prediction model mainly uses the yolov5s algorithm as a prediction network to predict targets and key points. Through the key pixel point prediction model, the adaptability and robustness of the prediction method can be effectively improved. In addition, in the present embodiment, in order to improve the accuracy and success rate of the prediction, the key pixel point prediction model predicts the output number of target key pixel points, and optionally, 72 key pixel points in the vehicle are used as the output of the model.

[0054] The image data can be input into a preset model to predict the target key pixel points of the target area in the vehicle. Since the target key pixel points have two-dimensional coordinates, the two-dimensional coordinates of the target key pixel points constitute a two-dimensional coordinate point set Q.

[0055] Among them, in the preset model (i.e., the target key pixel prediction model) provided in this embodiment, any key pixel prediction algorithm based on deep learning can be used, such as a hand gesture key pixel prediction algorithm, or a human body key pixel prediction algorithm based on the human body, or a face key pixel prediction algorithm based on the face. Of course, traditional corner detection algorithms can also be used, such as sift, surf, orb and other related key pixel prediction algorithms.

[0056] Furthermore, after obtaining the target key pixel points output by the key pixel prediction model, a two-dimensional coordinate point set Q of the key pixel points under the vehicle camera perspective is generated based on these two-dimensional key pixel points. However, since the model prediction will have some prediction errors to a greater or lesser extent, there is some deviation between the predicted key points and the actual positions. In order to reduce the prediction errors, an error correction algorithm is used in this embodiment, and the two-dimensional coordinate point set Q is error corrected to correct the prediction errors.

[0057] In this embodiment, the key pixels are first divided into different regions, and for each region, the minimum x coordinate, the minimum y coordinate, the maximum x coordinate, and the maximum y coordinate of the key pixels are calculated respectively.

[0058] Xmin=min(x0,x1,x2,x3…,xn,xn+1) (1)

[0059] Xmax=max(x0,x1,x2,x3…,xn,xn+1) (2)

[0060] Ymax=max(y0,y1,y2,y3,...yn,yn+1) (3)

[0061] Ymin=max(y0,y1,y2,y3,...yn,yn+1) (4)

[0062] The initial image area where the target key pixel point is located is determined by the above-mentioned two-dimensional coordinate point set Q. Then, the initial image area is subjected to image error correction processing to obtain the target image area. The error correction processing may include image grayscale processing, image edge detection processing, and image connected domain detection processing. The edge contour of the image can be obtained by the image edge detection algorithm, but since there is a large amount of noise in the image, in order to reduce the interference of noise points, the image connected domain detection analysis is performed to remove the noise contour area with a smaller area to obtain the true target image area, and the two-dimensional coordinate point set S after error correction is obtained based on the target image area and the two-dimensional coordinate point set Q.

[0063] in, Figure 2 is a schematic diagram of an image before error correction processing shown in an embodiment of the present application, such as Figure 2 As shown, Figure 2 Shown is an image before the above-mentioned image error correction processing is performed. Figure 3 is a schematic diagram of an image after error correction processing shown in an embodiment of the present application, such as Figure 3 As shown, Figure 3 What is shown is the image obtained after the above-mentioned image error correction processing.

[0064] In step 102, a three-dimensional coordinate point set P corresponding to the error-corrected two-dimensional coordinate point set S is obtained, and the external parameters of the camera are calibrated according to the two-dimensional coordinate point set S and the three-dimensional coordinate point set P.

[0065] In this embodiment, a three-dimensional coordinate system is established based on the vehicle model data. The location of each component in the vehicle cabin can be obtained in advance as the vehicle model data, and then the three-dimensional coordinate system is established based on the relative positions of each component. Alternatively, the three-dimensional data in the vehicle cabin can be collected by a three-dimensional sensor such as a laser radar, a three-dimensional scanner, or a depth camera to construct a three-dimensional coordinate system.

[0066] Among them, the three-dimensional coordinate point set P corresponding to the two-dimensional coordinate point set S after error correction can be obtained in the following manner: according to the vehicle model data of the vehicle, the vehicle parts where the key points corresponding to the target key pixel points are located are determined; the vehicle model data includes the three-dimensional coordinate position information of the points of the vehicle parts of the vehicle; according to the vehicle parts, the three-dimensional coordinate position information of the key points is determined; wherein the three-dimensional coordinate position information indicates the three-dimensional coordinates of the key points; according to each three-dimensional coordinate position information, the three-dimensional coordinate point set P is generated.

[0067] According to the target key pixel points obtained in step S101, the two-dimensional coordinates of each point in the target key pixel points are matched one by one with each three-dimensional coordinate in the three-dimensional coordinate system, thereby generating multiple coordinate pairs, and obtaining a three-dimensional coordinate point set P corresponding to the error-corrected two-dimensional coordinate point set S. According to the two-dimensional coordinate point set S and the three-dimensional coordinate point set P, the mapping relationship between the two-dimensional point set and the three-dimensional point set is determined using the PNP (Perspective-n-Point) algorithm, and the initial camera extrinsic parameters (R0, T0) of the camera inside the vehicle can be obtained.

[0068] The initial camera extrinsic parameters are iteratively processed, and when the preset conditions are met, the target camera extrinsic parameters are obtained.

[0069] For the target areas of all target key pixel points in the vehicle cabin, curve equations can be set, and these curve equations can be straight lines, two-dimensional curves, etc. Then, the target key pixel points of each area are brought into the curve equation, and the parameters of each curve equation are calculated to obtain the final curve.

[0070] In this embodiment, after the initial camera extrinsic parameters of the camera are obtained, the next iteration process is entered. In this iteration process, a curve fitting operation is performed on the error-corrected two-dimensional coordinate point set S to obtain at least one curve. The curve fitting method can use a straight line equation or a curve equation.

[0071] For at least one of the above curves, a point picking operation is performed according to a preset rule, wherein the preset rule may be that for each curve, a point picking operation is performed at intervals of one or several pixels (i.e., a point picking operation is performed using an interval sampling method). Through this dense sampling, a super-complete new two-dimensional coordinate point set H can be obtained, thereby enabling the optimization result of the camera extrinsic parameter to be more accurate. Alternatively, the preset rule may be a random sampling method to perform a point picking operation.

[0072] According to the initial camera extrinsic parameters, the three-dimensional coordinate point set can be remapped to the two-dimensional image to obtain the two-dimensional coordinate point set B corresponding to the three-dimensional coordinate point set. Then, the two-dimensional coordinate point set B and the super-complete two-dimensional coordinate point set H are compared one by one, and the three-dimensional coordinate point set P is mapped to the two-dimensional coordinate point set B through the camera extrinsic parameters calculated in the previous iteration. The point closest to each coordinate point in the two-dimensional coordinate point set B is found in the two-dimensional coordinate point set H, and the two-dimensional coordinate point set G is generated; the two-dimensional coordinate point set G and its corresponding three-dimensional coordinate point set P are used to re-update the camera extrinsic parameters obtained in the previous iterative processing process through the PNP algorithm (that is, the camera extrinsic parameters are updated using the two-dimensional coordinate point set G and the three-dimensional coordinate point set P). In this way, through repeated iterative processing, when the preset conditions are met, the iteration is stopped, and the camera extrinsic parameters obtained from the last iterative processing are used as the target camera extrinsic parameters to be obtained (Rn ,T n ).

[0073] Among them, the preset condition can be that when the change of the external parameter obtained by the current iterative process compared with the external parameter obtained by the previous iterative process is less than the preset rule, the iteration is stopped. The preset rule can be that the mapping relationship of all coordinate pairs in the two-dimensional coordinate point set and the three-dimensional coordinate point set does not change, or the mapping relationship of more than 95% of the coordinate pairs does not change. No specific limitation is made here, and the preset rule can be adjusted according to actual application needs. In another case, it can be set to stop the iteration when the number of iterations exceeds the preset threshold. In this embodiment, the specific number of iterations is not specifically limited.

[0074] In another embodiment, the above-mentioned camera extrinsic calibration method can be performed in a vehicle. Alternatively, since the computing power of the processor in the vehicle is relatively weak, in order to improve the speed and efficiency of calibration, after the image data in the vehicle is collected by the camera in the vehicle, the image data can be sent to a cloud server or other server that establishes a communication connection with the vehicle, so that the cloud server or other server that establishes a communication connection with the vehicle performs the camera's extrinsic calibration operation.

[0075] like Figure 4 As shown, Figure 4 It is a structural block diagram of a camera extrinsic parameter calibration device shown in an embodiment of the present application. The camera extrinsic parameter calibration device 400 includes: an acquisition module 401, used to acquire image data inside the vehicle collected by a camera inside the vehicle; an image processing module 402, used to obtain a two-dimensional coordinate point set Q corresponding to key pixel points of the target in the vehicle based on the image data, and perform error correction processing on the two-dimensional coordinate point set Q to obtain a two-dimensional coordinate point set S after error correction; a mapping module 403, used to acquire a three-dimensional coordinate point set P corresponding to the two-dimensional coordinate point set S after error correction, and calibrate the camera's extrinsic parameters according to the two-dimensional coordinate point set S and the three-dimensional coordinate point set P after error correction.

[0076] In another embodiment, the image processing module 402 is used to determine the initial image area where the target key pixel point is located based on the two-dimensional coordinate point set Q; perform image error correction processing on the initial image area to obtain the target image area, and obtain the corrected two-dimensional coordinate point set S based on the target image area and the two-dimensional coordinate point set Q.

[0077] In another embodiment, the image processing module 402 obtains target key pixel points of the target area in the vehicle based on the image data and a preset model prediction; and generates a two-dimensional coordinate point set Q according to the two-dimensional coordinates of the target key pixel points.

[0078] In another embodiment, the mapping module 403 is used to repeatedly perform the following steps until a preset condition is met: curve fitting is performed on the error-corrected two-dimensional coordinate point set S to obtain at least one curve;

[0079] For at least one curve, a point-taking operation is performed according to a preset rule to obtain a new two-dimensional coordinate point set H; wherein the preset rule is an interval sampling method or a random sampling method; the three-dimensional coordinate point set P is transformed by the camera external parameters calibrated last time to obtain a two-dimensional coordinate point set B, and the point closest to each coordinate point in the two-dimensional coordinate point set B is found in the new two-dimensional coordinate point set H, and a two-dimensional coordinate point set G is generated; the two-dimensional coordinate point set G and the three-dimensional coordinate point set P are used to update the external parameters of the camera; wherein the external parameters obtained when the preset conditions are met are the final external parameters determined.

[0080] In another embodiment, the preset condition is that the change of the external parameter obtained by the current iterative processing compared to the external parameter obtained by the previous iterative processing is less than a preset threshold; or, the preset condition is that the number of iterations exceeds the preset threshold.

[0081] In another embodiment, the mapping module 403 is used to determine the vehicle parts where the key points corresponding to the target key pixel points are located based on the vehicle model data of the vehicle; wherein the vehicle model data includes the three-dimensional coordinate position information of the points of the vehicle parts of the vehicle; according to the vehicle parts, the three-dimensional coordinate position information of the key points is determined; wherein the three-dimensional coordinate position information indicates the three-dimensional coordinates of the key points; and according to each three-dimensional coordinate position information, a three-dimensional coordinate point set P is generated.

[0082] In another embodiment, the acquisition module 401 is used to send the image data to a cloud server; wherein the image data is used to perform a calibration operation of the camera's external parameters.

[0083] The present disclosure also provides a vehicle, Figure 5 is a structural block diagram of a vehicle shown in an embodiment of the present application, such as Figure 5 As shown, the vehicle includes: a camera, at least one processor, and a memory communicatively connected to the at least one processor; wherein the camera is used to collect image data within the vehicle; the memory stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor so that the at least one processor can execute the camera extrinsic parameter calibration method of an embodiment of the present application.

[0084] The disclosed embodiment also provides a computer-readable storage medium, which has computer instructions. When the computer instructions are executed, the camera extrinsic parameter calibration algorithm of the embodiment of the present application is implemented.

[0085] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this specification. A person of ordinary skill in the art can understand and implement it without paying creative labor.

[0086] The above description is of certain embodiments of the present invention. Other embodiments are within the scope of the appended claims.

[0087] In some cases, the actions or steps described in the claims may be performed in a different order than that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0088] Those skilled in the art will readily appreciate other embodiments of the specification after considering the specification and practicing the invention claimed herein. The specification is intended to cover any variations, uses or adaptations of the specification that follow the general principles of the specification and include common knowledge or customary techniques in the art that are not claimed in the specification. The specification and examples are to be considered exemplary only, and the true scope and spirit of the specification are indicated by the following claims.

[0089] It should be understood that the present description is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present description is limited only by the appended claims.

[0090] The above description is only a preferred embodiment of this specification and is not intended to limit this specification.

[0091] Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles shall be included in the scope of protection of this manual.

Claims

1. A camera extrinsic calibration method, It is characterized in that The method comprises: Acquire image data inside the vehicle collected by a camera inside the vehicle; Based on the image data, a two-dimensional coordinate point set Q corresponding to a target key pixel point in the vehicle is obtained, and error correction processing is performed on the two-dimensional coordinate point set Q to obtain a two-dimensional coordinate point set S after error correction; A three-dimensional coordinate point set P corresponding to the error-corrected two-dimensional coordinate point set S is obtained, and the external parameters of the camera are calibrated according to the error-corrected two-dimensional coordinate point set S and the three-dimensional coordinate point set P.

2. The method according to claim 1, It is characterized in that Performing error correction processing on the two-dimensional coordinate point set Q to obtain a two-dimensional coordinate point set S after error correction, including: Determine the initial image area where the target key pixel point is located according to the two-dimensional coordinate point set Q; The initial image region is subjected to image error correction processing to obtain a target image region, and the error-corrected two-dimensional coordinate point set S is obtained according to the target image region and the two-dimensional coordinate point set Q.

3. The method according to claim 1, It is characterized in that The step of obtaining a two-dimensional coordinate point set Q of key pixel points of a target in the vehicle based on the image data includes: Obtaining target key pixel points of the target area in the vehicle based on the image data and a preset model prediction; The two-dimensional coordinate point set Q is generated according to the two-dimensional coordinates of the target key pixel points.

4. The method according to claim 1, It is characterized in that Calibrate the external parameters of the camera according to the corrected two-dimensional coordinate point set S and the three-dimensional coordinate point set P, including: Repeat the following steps until the preset conditions are met: Performing curve fitting on the error-corrected two-dimensional coordinate point set S to obtain at least one curve; For the at least one curve, a point sampling operation is performed according to a preset rule to obtain a new two-dimensional coordinate point set H; wherein the preset rule is an interval sampling method or a random sampling method; The three-dimensional coordinate point set P is transformed by the camera external parameters calibrated last time to obtain a two-dimensional coordinate point set B, and the point closest to each coordinate point in the two-dimensional coordinate point set B is found in the new two-dimensional coordinate point set H, and a two-dimensional coordinate point set G is generated; Using the two-dimensional coordinate point set G and the three-dimensional coordinate point set P, updating the external parameters of the camera; Among them, the external parameters obtained when the preset conditions are met are the final external parameters determined.

5. The method according to claim 4, It is characterized in that The preset condition is When the change of the external parameter obtained by the current iterative processing compared with the external parameter obtained by the previous iterative processing is less than a preset threshold; Alternatively, the preset condition is that the number of iterations exceeds a preset threshold.

6. The method according to any one of claims 1 to 5, It is characterized in that Acquiring a three-dimensional coordinate point set P corresponding to the error-corrected two-dimensional coordinate point set S, comprising: According to the vehicle model data of the vehicle, determine the vehicle component where the key point corresponding to the target key pixel point is located; wherein the vehicle model data includes the three-dimensional coordinate position information of the point of the vehicle component of the vehicle; Determine the three-dimensional coordinate position information of the key point according to the vehicle component; wherein the three-dimensional coordinate position information indicates the three-dimensional coordinates of the key point; The three-dimensional coordinate point set P is generated according to each of the three-dimensional coordinate position information.

7. The method according to any one of claims 1 to 5, It is characterized in that The method further comprises: The image data is sent to a cloud server; wherein the image data is used to perform a calibration operation of the camera's external parameters.

8. A camera extrinsic calibration device, It is characterized in that The device comprises: An acquisition module, used for acquiring image data inside the vehicle captured by a camera inside the vehicle; An image processing module, used for obtaining a two-dimensional coordinate point set Q corresponding to the target key pixel points in the vehicle based on the image data, and performing error correction processing on the two-dimensional coordinate point set Q to obtain a two-dimensional coordinate point set S after error correction; A mapping module is used to obtain a three-dimensional coordinate point set P corresponding to the error-corrected two-dimensional coordinate point set S, and calibrate the external parameters of the camera according to the error-corrected two-dimensional coordinate point set S and the three-dimensional coordinate point set P.

9. A vehicle, It is characterized in that The vehicle comprises: a camera, at least one processor, and a memory in communication with the at least one processor; wherein, The camera is used to collect image data inside the vehicle; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium has computer instructions, and when the computer instructions are executed, the method according to any one of claims 1 to 7 is implemented.

Images