Parameter evaluation method, device and electronic device

Abstract

The present invention provides a parameter evaluation method, device and electronic device, which obtain a first edge map and a second edge map of a target scene; wherein the first edge map is obtained by a camera device; the second edge map is obtained by a laser radar; multiple groups of parameters are obtained to determine the correlation between the first edge map and the second edge map corresponding to each group of parameters; wherein the multiple groups of parameters include current parameters and multiple groups of perturbation parameters obtained by performing perturbation analysis on the current parameters; the parameters are used to indicate the posture relationship between the camera device and the laser radar; and the evaluation result of the current parameter is determined according to the correlation corresponding to each group of parameters. In this method, multiple groups of parameters are obtained through perturbation analysis, and the corresponding correlation is determined based on each group of parameters, so that the accuracy of the current parameter can be determined in real time. Compared with the evaluation and marking method in an offline state, the accuracy of the current parameter is evaluated in real time, thereby improving the accuracy of the perception data.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a parameter evaluation method, device and electronic equipment. Background Art

[0002] In order to adapt to accurate perception in complex environments, smart cars, robots and other devices usually need to combine cameras and lidar sensors to perceive data in complex environments; in order to make the perception data more standardized, it is necessary to evaluate the accuracy of the parameters between the camera and the lidar. In related technologies, the parameters between the camera and the lidar are usually calibrated offline using reflective markers, chessboards or manual markings; however, in actual use, if the relative position between the camera and the lidar changes, the parameters cannot be detected and evaluated in real time; at this time, if the calibration results in the offline state are still applied to the currently running device, the accuracy of the perception data will be reduced. Summary of the invention

[0003] In view of this, an object of the present invention is to provide a parameter evaluation method, device and electronic device to detect and evaluate the parameters between the camera device and the laser radar in real time, thereby improving the accuracy of the perception data.

[0004] In a first aspect, an embodiment of the present invention provides a parameter evaluation method, comprising: obtaining a first edge map and a second edge map of a target scene; wherein the first edge map is obtained by a camera device; and the second edge map is obtained by a laser radar; obtaining multiple sets of parameters, and determining the correlation between the first edge map and the second edge map corresponding to each set of parameters; the multiple sets of parameters include current parameters, and multiple sets of perturbation parameters obtained by performing perturbation analysis on the current parameters; the parameters are used to indicate the posture relationship between the camera device and the laser radar; and determining the evaluation result of the current parameter according to the correlation corresponding to each set of parameters.

[0005] Furthermore, the step of obtaining multiple groups of parameters and determining the correlation between the first edge map and the second edge map corresponding to each group of parameters includes: for each group of parameters, projecting the second edge map onto the first edge map through the parameters, removing the points of the second edge map outside the first edge map, and obtaining a third edge map corresponding to the second edge map; and calculating the correlation between the first edge map and the second edge map corresponding to each group of parameters based on the second edge map and the third edge map corresponding to the second edge map.

[0006] Furthermore, the second edge map includes multiple images; the step of calculating the correlation between the first edge map and the second edge map corresponding to each set of parameters according to the second edge map and the third edge map corresponding to the second edge map includes: for each set of parameters, calculating the product value of the depth value of each pixel point in each second edge map and the depth value of the pixel point at the corresponding position in the third edge map corresponding to each second edge map; and determining the sum of the product values ​​corresponding to each second edge map as the correlation between the first edge map and the second edge map corresponding to the parameter.

[0007] Furthermore, the correlation degree includes: the correlation degree corresponding to the current parameter and the correlation degree corresponding to the disturbance parameter; the step of determining the evaluation result of the current parameter according to the correlation degree corresponding to each group of parameters includes: obtaining the target correlation degree from the correlation degree corresponding to the disturbance parameter; wherein the target correlation degree is less than the correlation degree corresponding to the current parameter; calculating the ratio of the number of target correlation degrees to the number of correlation degrees corresponding to the disturbance parameter; and determining the ratio as the evaluation result of the current parameter.

[0008] Furthermore, the step of obtaining a first edge map and a second edge map of the target scene includes: capturing an original image of the target scene through a camera device and converting the original image into a grayscale image; obtaining the pixel point with the largest grayscale value in each specified area of ​​the grayscale image in a preset order; determining an edge map of the grayscale image based on the pixel point with the largest grayscale value; and performing an inverse distance transform on the edge map to obtain a first edge map.

[0009] Furthermore, the step of obtaining a first edge map and a second edge map of the target scene includes: collecting point cloud data of the target scene through a laser radar; filtering out noise points in the point cloud data according to the depth value of each position point in the point cloud data to obtain filtered point cloud data; obtaining edge points in the filtered point cloud data through the depth continuity between each position point in the filtered point cloud data and adjacent position points; and determining a second edge map based on the edge points.

[0010] Furthermore, according to the depth value of each position point in the point cloud data, the noise points in the point cloud data are filtered out to obtain the filtered point cloud data, including: for each position point in the point cloud data, calculating the absolute value of the first depth difference between the depth value of the position point and the depth value of the previous position point, and the absolute value of the second depth difference between the depth value of the position point and the depth value of the next position point; if the absolute value of the first depth difference and the absolute value of the second depth difference are both greater than a first specified depth threshold, filtering the position point to obtain the filtered point cloud data.

[0011] Furthermore, edge points in the filtered point cloud data are obtained by determining the depth continuity between each position point in the filtered point cloud data and adjacent position points; based on the edge points, the step of determining the second edge map includes: for each position point in the filtered point cloud data, calculating the absolute value of the third depth difference between the depth value of the position point and the depth value of the previous position point, and the absolute value of the fourth depth difference between the depth value of the position point and the depth value of the next position point; obtaining the maximum value of the absolute value of the third depth difference and the absolute value of the fourth depth difference; if the maximum value is greater than the second specified depth threshold, determining the position point as an edge point; and determining the second edge map based on the edge point.

[0012] Furthermore, the evaluation result includes an evaluation index; after the step of determining the evaluation result of the current parameter according to the correlation corresponding to each group of parameters, the method also includes: calculating the accuracy of the current parameter based on the mean and standard deviation of the evaluation index corresponding to the preset correct parameter, and the mean and standard deviation of the evaluation index corresponding to the preset wrong parameter; if the accuracy of the current parameter is less than a specified threshold, obtaining multiple groups of specified parameters by performing a perturbation analysis of the current parameter with a specified perturbation radius; calibrating the current parameter based on the correlation corresponding to the current parameter and the correlation corresponding to each group of specified parameters.

[0013] Furthermore, based on the correlation degree corresponding to the current parameter and the correlation degree corresponding to each group of specified parameters, the current step is calibrated, including: for the correlation degree corresponding to each group of specified parameters, multiple evaluation indicators under the specified parameters are calculated; the mean and standard deviation of the multiple evaluation indicators are calculated, and the specified parameters that meet the preset mean and preset standard deviation are determined as target parameters; and the target parameters are calibrated to the current parameters.

[0014] In a second aspect, an embodiment of the present invention provides a parameter evaluation device, comprising: an acquisition module, used to acquire a first edge map and a second edge map of a target scene; wherein the first edge map is obtained by a camera device; and the second edge map is obtained by a laser radar; a determination module, used to determine, based on multiple groups of parameters, a correlation between the first edge map and the second edge map corresponding to each group of parameters; wherein the multiple groups of parameters include current parameters and multiple groups of disturbance parameters obtained by performing perturbation analysis on the current parameters; and an evaluation module, used to determine an evaluation result of the current parameter according to the correlation corresponding to each group of parameters.

[0015] In a third aspect, an embodiment of the present invention provides an electronic device, including a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the parameter evaluation method of any one of the first aspects.

[0016] In a fourth aspect, an embodiment of the present invention provides a machine-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the parameter evaluation method of any one of the first aspects.

[0017] The embodiments of the present invention bring the following beneficial effects:

[0018] The present invention provides a parameter evaluation method, device and electronic device, which obtain a first edge map and a second edge map of a target scene; wherein the first edge map is obtained by a camera device; the second edge map is obtained by a laser radar; multiple groups of parameters are obtained to determine the correlation between the first edge map and the second edge map corresponding to each group of parameters; wherein the multiple groups of parameters include current parameters and multiple groups of perturbation parameters obtained by performing perturbation analysis on the current parameters; the parameters are used to indicate the posture relationship between the camera device and the laser radar; and the evaluation result of the current parameter is determined according to the correlation corresponding to each group of parameters. In this method, multiple groups of parameters are obtained through perturbation analysis, and the corresponding correlation is determined based on each group of parameters, so that the accuracy of the current parameter can be determined in real time. Compared with the evaluation and marking method in an offline state, the accuracy of the current parameter is evaluated in real time, thereby improving the accuracy of the perception data.

[0019] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 A flow chart of a parameter evaluation method provided by an embodiment of the present invention;

[0023] Figure 2 A flowchart of another parameter evaluation method provided by an embodiment of the present invention;

[0024] Figure 3A flow chart of a parameter calibration method provided by an embodiment of the present invention;

[0025] Figure 4 A schematic diagram of the structure of a parameter evaluation device provided by an embodiment of the present invention;

[0026] Figure 5 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0028] In order to adapt to accurate perception in complex environments, the number and form of sensors configured in devices such as smart cars or robots are gradually increasing. Compared with devices with a single sensor, higher-level tasks can fuse multiple sensors to perform perception tasks, among which multi-sensor systems can be directly applied to three-dimensional environment scanning. For example, by combining cameras and lidars, lidars can directly provide accurate distance measurement information, and cameras can provide high spatial resolution and valuable color information. Through the complementary information of the two sensors, the distance image can be color mapped, and each colored point in the lidar point cloud can be used to build a richer regional model. Therefore, accurate calibration between lidars and cameras is particularly important.

[0029] However, due to the different forms of sensors and the nature of output information, sensor extrinsic parameter calibration is quite challenging. In related technologies, sensor extrinsic parameters are usually calibrated by manual calibration. Automatic calibration is very important for mobile robots or smart cars that rely on multi-sensor modes. The current common method is to use a manual method to perform a calibration once in an offline state, and assume that this calibration remains unchanged for a period of time. However, these methods are only applicable to a very small range of situations and laser radar types. This offline algorithm lacks generalization ability. During actual operation, if the relative position between the camera and the laser radar changes, the parameters cannot be detected and evaluated in real time; at this time, if the calibration results in the offline state are still applied to the currently running device, the accuracy of the perception data will be reduced. Based on this, an embodiment of the present invention provides a parameter evaluation method, device and electronic device, which can be applied to devices with perception functions, especially to devices such as smart cars or robots.

[0030] To facilitate understanding of this embodiment, a parameter evaluation method disclosed in an embodiment of the present invention is first described in detail. The method is applied to an electronic device, which may be a smart car, a robot, etc. Figure 1 As shown, the method comprises the following steps:

[0031] Step S102, obtaining a first edge map and a second edge map of the target scene; wherein the first edge map is obtained by a camera device; and the second edge map is obtained by a laser radar;

[0032] The target scene may be a scene that the device needs to perceive during operation, such as objects such as roads and shelves. The first edge map may be obtained by performing image processing using an image captured by a camera; the second edge map may be obtained by performing data processing using point cloud data collected by a laser radar. Specifically, during the operation of a device including a camera and a laser radar, the first edge map and the second edge map of the target scene are acquired in real time based on the current camera and the laser radar.

[0033] Step S104, obtaining multiple sets of parameters, and determining the correlation between the first edge map and the second edge map corresponding to each set of parameters; the multiple sets of parameters include current parameters, and multiple sets of perturbation parameters obtained by performing perturbation analysis on the current parameters; the parameters are used to indicate the posture relationship between the camera device and the laser radar;

[0034] The above parameters usually include a rotation matrix and a translation matrix, which can also be called an external parameter, and is used to indicate the positional relationship between the camera and the laser radar, that is, the image collected by the camera can be fused with the point cloud data collected by the laser radar through this parameter. Usually, the device will calibrate the parameters between the camera and the laser radar before starting operation, that is, the above current parameters. During the actual operation process, or before the operation, multiple sets of parameters can be obtained based on the current parameters. Specifically, multiple sets of perturbation parameters can be obtained by performing perturbation analysis on the current parameters. For example, the x degrees of freedom of the current parameters are perturbed by a small amount in the positive and negative directions respectively. Each degree of freedom can generate three sets of parameters, namely the positive direction, the negative direction, and the current position, for a total of 3 x Group parameters.

[0035] The above correlation degree is used to indicate the degree of alignment between the second edge map of the current laser radar point and the first edge map. The degree of alignment can be the product value or the added value of each aligned pixel point. In actual implementation, the first edge map actually collected by the running camera device and the second edge map actually collected by the running laser radar can be used. Since the first edge map and the second edge map are collected at the same time for the same target scene, the first edge map and the second edge map correspond to each other, and the corresponding relationship can be converted by the above parameters; specifically, the first edge map and the second edge map can be converted by the above multiple sets of parameters, the first edge map and the second edge map are aligned, and the processed image is used to calculate the correlation degree between the first edge map and the second edge map corresponding to the set of parameters.

[0036] It is understandable that each set of parameters is different, and the alignment process of the first edge map and the second edge map based on each set of parameters may result in some deviation in the obtained result.

[0037] Step S106, determining the evaluation result of the current parameter according to the correlation degree corresponding to each group of parameters.

[0038] The above evaluation results may include the accuracy of the current parameters. Specifically, according to step S104, the correlation degree corresponding to the current parameters and the correlation degree corresponding to the disturbance parameters can be obtained. It can be understood that if the current parameters are accurate parameters, in actual operation, the position between the camera device and the laser radar has not changed, then it can be determined that the correlation degree corresponding to the current parameters is the largest, that is, the degree of alignment is large, and at this time it can be determined that the current parameters are accurate parameters. If the current parameters are accurate parameters, in actual operation, the position between the camera device and the laser radar has changed to a certain extent, then it can be determined that the correlation degree corresponding to the current parameters is not the largest among the correlation degrees corresponding to all disturbance external parameters; at this time it can be determined that the current parameters are not accurate parameters.

[0039] Specifically, the correlation degree corresponding to the current parameter can be compared with the correlation degrees corresponding to all disturbance parameters. If the correlation degree corresponding to the current parameter is greater than the correlation degrees corresponding to a specified number of disturbance parameters, for example, the specified number can be more than 90% of the number of disturbance parameters.

[0040] The present invention provides a parameter evaluation method, which obtains a first edge map and a second edge map of a target scene; wherein the first edge map is obtained by a camera device; the second edge map is obtained by a laser radar; multiple groups of parameters are obtained to determine the correlation between the first edge map and the second edge map corresponding to each group of parameters; wherein the multiple groups of parameters include current parameters, and multiple groups of perturbation parameters obtained by performing perturbation analysis on the current parameters; the parameters are used to indicate the posture relationship between the camera device and the laser radar; and the evaluation result of the current parameter is determined according to the correlation corresponding to each group of parameters. In this method, multiple groups of parameters are obtained through perturbation analysis, and the corresponding correlation is determined based on each group of parameters, so that the accuracy of the current parameter can be determined in real time. Compared with the evaluation and marking method in an offline state, the accuracy of the current parameter can be evaluated in real time, thereby improving the accuracy of the perception data.

[0041] This embodiment also provides another parameter evaluation method, which is implemented on the basis of the above embodiment. This embodiment focuses on the specific implementation process of the steps of obtaining multiple groups of parameters and determining the correlation between the first edge map and the second edge map corresponding to each group of parameters (implemented by steps S204-S206), and the specific implementation process of the step of determining the evaluation result of the current parameter according to the correlation corresponding to each group of parameters (implemented by steps S208-S210), such as Figure 2 As shown, the method comprises the following steps:

[0042] Step S202, obtaining a first edge map and a second edge map of the target scene; wherein the first edge map is obtained by a camera device; and the second edge map is obtained by a laser radar;

[0043] Step S204, obtaining multiple sets of parameters, and for each set of parameters, projecting the second edge map onto the first edge map through the parameters, removing points of the second edge map outside the first edge map, and obtaining a third edge map corresponding to the second edge map;

[0044] For example, the 6 degrees of freedom of the current parameter are perturbed in small amounts in the positive and negative directions, and 3 6= 729 groups of parameters, one of which is the current parameters, and the other 728 groups of parameters are perturbation parameters; for each group of parameters in the 729 groups of parameters, firstly, the second edge map is projected onto the first edge map through the current parameters. Specifically, each pixel point in the second edge map is projected onto the first edge map using the rotation matrix and translation matrix in the current parameters. Due to the inaccuracy of the current parameters, a certain deviation may occur, or the second edge map may be too large, and redundant pixels may be generated outside the first edge map. At this time, the points of the second edge map outside the first edge map need to be removed, and the pixels projected onto the first edge map are determined as the third edge map. Then, the above steps are performed for other perturbation parameters to obtain the third edge map of each group of parameters corresponding to the second image.

[0045] Step S206 , calculating the degree of association between the first edge map and the second edge map corresponding to each group of parameters according to the second edge map and the third edge map corresponding to the second edge map.

[0046] Specifically, the degree of matching between the pixel value of each pixel point in the third edge map corresponding to the second edge map and the pixel value of the corresponding pixel point in the second edge map can be calculated. For example, the difference between the pixel value of each pixel point in the third edge map corresponding to the second edge map and the pixel value of the corresponding pixel point in the second edge map can be calculated. If the difference is smaller, it means that the correlation between the first edge map and the second edge map is higher.

[0047] For another example, the product of the pixel value of each pixel position in the third edge map corresponding to the second edge map and the pixel value of the corresponding pixel position in the second edge map is calculated. If the product value is larger, it means that the correlation between the first edge map and the second edge map is higher. It can be understood that for a pixel point that does not correspond to a pixel position, the pixel value of the pixel position in the third edge map is zero, or a smaller value. At this time, the product of the pixel value of the pixel position on the second edge map and the pixel value of the pixel position on the third edge map is much smaller than the product of the pixel values ​​of the pixel points aligned with the pixel position.

[0048] Furthermore, the second edge map includes multiple images; that is, the target scene can be continuously photographed in actual operation to obtain multiple first edge maps and second edge maps; a possible implementation method:

[0049] (1) For each set of parameters, calculate the product of the depth value of each pixel in each second edge map and the depth value of the pixel at the corresponding position in the third edge map corresponding to each second edge map;

[0050] (2) The sum of the product values ​​corresponding to each second edge map is determined as the correlation degree between the first edge map and the second edge map corresponding to the parameter.

[0051] In order to increase the credibility of the correlation, calculations can be performed for multiple first edge images and second edge images. Specifically, when the device is running, multiple first edge images and second edge images can be obtained within a currently specified period of time, and the images obtained within the currently specified period of time can be saved as the current data set. Specifically, the correlation corresponding to each set of parameters can be calculated by the following formula:

[0052]

[0053] In the above formula, n Indicates the total number of image frames in the current dataset; w Indicates the number of frames closest to the current frame; m Indicates the frame number of the starting frame of the current data set; Represents the total number of pixels in the second edge map, p Represents the index of the pixel on the second edge map; Indicates that the index on the second edge graph is p The depth value of the pixel point; Represents the depth value of the pixel at position i, j on the third edge map, where i, j Represents the second edge graph with index as p The pixel point corresponds to the pixel position on the third edge map.

[0054] Specifically, for each second edge map, a third edge map corresponding to the second edge map is obtained; then, for each second edge map, a depth value of each pixel position in the second edge map is calculated, and the product of the depth value of the pixel position in the third edge map corresponding to the pixel position is calculated, and the product values ​​corresponding to each second edge map are added to obtain the correlation degree corresponding to each group of parameters.

[0055] The above correlation degree includes: the correlation degree corresponding to the current parameter and the correlation degree corresponding to the disturbance parameter;

[0056] Step S208, obtaining a target correlation degree from the correlation degrees corresponding to the disturbance parameters; wherein the target correlation degree is less than the correlation degree corresponding to the current parameters;

[0057] Step S210, calculating the ratio of the number of target association degrees to the number of association degrees corresponding to the disturbance parameter; and determining the ratio as the evaluation result of the current parameter.

[0058] The above target correlation degree usually includes multiple ones. Taking the multiple groups of parameters including 729 groups of parameters as an example, the correlation degree corresponding to 728 groups of disturbance parameters is compared with the correlation degree corresponding to the current parameters, and the target correlation degree that is less than the correlation degree corresponding to the current parameters is obtained from the correlation degrees corresponding to the 728 groups of disturbance parameters; the number of target correlation degrees is counted, and the ratio of the number of target correlation degrees to the number of correlation degrees corresponding to the disturbance parameters is determined as the evaluation result of the current parameters; the specific formula is as follows:

[0059]

[0060] In the above formula, Indicates the number of target relevance; represents the number of correlations corresponding to the disturbance parameter (or the number of disturbance parameters); Represents the evaluation result of the current parameters, where if If it is close to 1 (100%), it means that the current parameters are correct. If it is significantly less than 1, it means that the current parameters are inaccurate.

[0061] In the above method, the accuracy of the external parameters of the camera device and the lidar can be automatically evaluated in the running state, without manual marking, complex devices and calibration processes, and fixed signs or feature scenes. Compared with the offline calibration method, it is more flexible and universal, and the accuracy of the current parameters can be evaluated in real time. Through real-time evaluation, an alarm notification can be issued, which consumes very little CPU time and has a fast response speed. In addition, compared with directly solving nonlinear problems, it saves more time and computing resources, greatly improving the evaluation efficiency.

[0062] This embodiment mainly describes the steps of obtaining the first edge map and the second edge map of the target scene. Specifically, the steps of obtaining the first edge map of the target scene include the following steps:

[0063] Step A1, collecting an original image of a target scene through a camera device, and converting the original image into a grayscale image; obtaining the pixel point with the maximum grayscale value in each specified area of ​​the grayscale image in a preset order; and determining an edge map of the grayscale image based on the pixel point with the maximum grayscale value;

[0064] The above-mentioned designated area can be a 3×3 area centered on the target pixel point. Specifically, for the pixel points in the grayscale image of the original image, a target pixel point is selected for each designated pixel point in the order from left to right and from top to bottom, and the pixel point with the maximum grayscale value is obtained in the designated area of ​​the target pixel point, and the pixel point with the maximum grayscale value obtained in each designated area of ​​the grayscale image is determined as the edge map of the grayscale image.

[0065] Taking the size of the designated area as 3×3 as an example, from top to bottom, a target pixel is selected every 3 pixels, and the pixel with the largest grayscale value is obtained in the 3×3 area of ​​the target pixel. The pixel with the largest grayscale value obtained in each 3×3 area of ​​the grayscale image is determined as the edge map of the grayscale image.

[0066] Step A2: Perform inverse distance transformation on the edge map to obtain a first edge map.

[0067] Specifically, the edge map can be processed by the following formula:

[0068]

[0069] In the above formula, Indicates that the coordinates in the first edge graph are i, j The gray value of the pixel; It means that the coordinates in the above edge graph are x,y The gray value of the pixel; i, j Represents the coordinates of the current pixel position in the above edge map; x,y Represents the coordinates of the pixel with the largest grayscale value in the neighborhood of the current pixel position in the above edge map, where Represents the neighborhood of the current pixel position; Indicates the weight of the gray value of the current pixel position, usually =1 / 3; represents the exponential smoothing factor, usually =0.98, It can make the edges of the image smoother.

[0070] In addition, the specific step of obtaining the second edge map of the target scene includes the following steps:

[0071] Step B1, collecting point cloud data of the target scene through a laser radar;

[0072] Step B2, filtering out noise points in the point cloud data according to the depth value of each position point in the point cloud data to obtain filtered point cloud data;

[0073] Specifically, the depth value of each frame of the point cloud data can be used to filter out noise points in the point cloud data through a noise reduction method to obtain filtered point cloud data, such as smoothing filtering, singular point removal, and other methods.

[0074] A possible implementation:

[0075] For each position point in the point cloud data, calculate the absolute value of the first depth difference between the depth value of the position point and the depth value of the previous position point, and the absolute value of the second depth difference between the depth value of the position point and the depth value of the next position point; if the absolute value of the first depth difference and the absolute value of the second depth difference are both greater than the first specified depth threshold, filter out the position point to obtain filtered point cloud data.

[0076] The first specified depth threshold can be set according to actual needs, such as 20, 30, etc. Generally, if the absolute value of the first depth difference and the absolute value of the second depth difference are both greater than the first specified depth threshold, it means that the position point is a noise point, and the point is excluded from the following calculation.

[0077] Step B3, obtaining edge points in the filtered point cloud data through depth continuity between each position point in the filtered point cloud data and adjacent position points; and determining a second edge map based on the edge points.

[0078] Specifically, the difference between the depth value of each position point and the depth value of the adjacent position point may be calculated, and whether there is continuity may be determined according to the difference, and the second edge map may be determined through the points with continuity.

[0079] A possible implementation:

[0080] The step of determining a second edge map based on edge points includes: for each position point in the filtered point cloud data, calculating an absolute value of a third depth difference between a depth value of the position point and a depth value of a previous position point, and an absolute value of a fourth depth difference between the depth value of the position point and a depth value of a subsequent position point; obtaining a maximum value between the absolute value of the third depth difference and the absolute value of the fourth depth difference; if the maximum value is greater than a second specified depth threshold, determining the position point as an edge point; and determining a second edge map based on the edge point.

[0081] Each position point in the above point cloud data can be represented by the index of the pixel point in each frame of the point cloud data. In actual implementation, the maximum value of the absolute value of the third depth difference and the absolute value of the fourth depth difference in the filtered point cloud data can be obtained by the following formula:

[0082]

[0083] In the above formula, represents the maximum value of the absolute value of the third depth difference and the absolute value of the fourth depth difference; i Indicates the laser radar level i Frame point cloud data; Indicates i The index in the frame point cloud data is p The depth value of the point, Indicates i The index in the frame point cloud data is p The depth value of the point is -1, Indicates i The index in the frame point cloud data is p +1 is the depth value of the point, Represents the depth continuity evaluation factor, usually =0.5 can make the edges of the image smoother.

[0084] In the above formula Can be expressed p The absolute value of the third depth difference between the depth value of the position point and the depth value of the previous position point; Can be expressed p The absolute value of the fourth depth difference between the depth value of the position point and the depth value of the next position point; the maximum value of the absolute value of the third depth difference and the absolute value of the fourth depth difference is obtained through max Usually when When it is greater than 30cm, the index is determined to be p The points are edge points of the second edge graph.

[0085] This embodiment also provides a parameter calibration method, which is implemented on the basis of the above embodiment. The above evaluation result includes evaluation indicators. This embodiment mainly describes the specific steps after the step of determining the evaluation result of the current parameter according to the correlation degree corresponding to each group of parameters, such as Figure 3 As shown, the method comprises the following steps:

[0086] Step S302, calculating the accuracy of the current parameters based on the mean and standard deviation of the evaluation indicators corresponding to the preset correct parameters and the mean and standard deviation of the evaluation indicators corresponding to the preset incorrect parameters;

[0087] The above preset correct parameters and the mean and standard deviation of the evaluation indicators corresponding to the errors can be determined in the following way:

[0088] For the same set of camera devices and laser radars, before actual operation, a large amount of point cloud data and image data can be collected based on correct parameters (the sensor posture can be adjusted) and time synchronization matching can be performed to obtain multiple first edge maps and second edge maps. Specifically, multiple first edge maps and second edge maps are grouped, and each group corresponds to a set of correct parameters (which can be obtained through offline calibration). Wrong parameters can be generated through random perturbations, and the F values ​​corresponding to multiple sets of correct parameters are calculated respectively. C Values, and F corresponding to multiple sets of error parameters C Then, the two parameters are statistically analyzed for F Cdistribution, namely the mean and standard deviation.

[0089] Specifically, the accuracy of the current parameters can be calculated by the following formula:

[0090]

[0091] In the above formula, x represents the accuracy evaluation index corresponding to the current parameter (and Fc obtained in step S210 above); and They represent the mean and standard deviation of the distribution of the accuracy evaluation index corresponding to the correct parameters, and They represent the mean and standard deviation of the distribution of the accuracy evaluation index corresponding to the error parameter respectively; Indicates the accuracy of the current parameters.

[0092] Step S304, if the accuracy of the current parameter is less than the specified threshold, a plurality of groups of specified parameters are obtained by performing a disturbance analysis of the current parameter with a specified disturbance radius;

[0093] Step S306: calibrate the current parameter based on the correlation degree corresponding to the current parameter and the correlation degree corresponding to each group of designated parameters.

[0094] The above specified threshold can be set according to actual needs. For example, if the specified threshold is 0.5, When it is greater than 0.5, it means the current parameters are correct. When it is less than 0.5, it means that the current parameter is invalid. The above-mentioned specified disturbance radius is usually larger than the disturbance radius of the disturbance analysis performed when evaluating the current parameter; the above-mentioned specified parameter can also be called a disturbance parameter. If the accuracy of the current parameter is less than the specified threshold, the current parameter can be calibrated. Specifically, the evaluation result corresponding to each group of specified parameters, that is, the accuracy evaluation index of each group of specified parameters, can be calculated, and the parameter with the largest accuracy evaluation index can be determined as the current parameter; the correlation degree corresponding to each group of specified parameters can also be calculated, and the parameter with the largest correlation degree can be determined as the current parameter, etc.

[0095] A possible implementation:

[0096] For the correlation degree corresponding to each set of specified parameters, multiple evaluation indicators under the specified parameters are calculated; the mean and standard deviation of the multiple evaluation indicators are calculated, and the specified parameters that meet the preset mean and preset standard deviation are determined as target parameters; the target parameters are calibrated to the current parameters.

[0097] Specifically, for each set of specified parameters, multiple sets of first edge maps and second edge maps can be obtained, and for each set of images, the correlation degree corresponding to the specified parameters can be calculated, that is, each set of specified parameters can obtain multiple corresponding correlation degrees; multiple evaluation indicators under the specified parameters are calculated based on each correlation degree. Then the mean and standard deviation of the multiple evaluation indicators are calculated, and finally the mean and standard deviation of the evaluation indicators corresponding to the multiple specified parameters are obtained; from the mean and standard deviation of the evaluation indicators corresponding to the multiple specified parameters, the specified parameters corresponding to the evaluation indicators with the mean and standard deviation corresponding to the preset correct parameters are obtained, and determined as the target parameters; the target parameters are calibrated to the current parameters.

[0098] It should be noted that if the above method does not obtain valid and correct parameters, the current parameters can be calibrated in the following way:

[0099] Method 1: Use the greedy algorithm to select the accuracy evaluation index F from the 729 sets of perturbation parameters generated by the current parameters C The maximum parameter value is used as a new starting point, and the process of step S304 to step S306 is continuously repeated until accurate parameters are obtained.

[0100] Method 2: A search method combining probability sampling and greedy algorithm. For the 729 sets of perturbation parameters generated by the current parameters, the accuracy evaluation index F C The probability distribution of the value is generated, for example, a roulette method is used to randomly sample a set of parameter values ​​from the generated probability distribution, and then the parameter candidate values ​​are found by performing a greedy algorithm search on the sampled parameter values, and the accuracy evaluation index F is selected from the finally found parameter candidate values. C Maximum value, accuracy evaluation index F C The correct external parameter can be obtained by averaging or taking other evaluation criteria. It is also possible to consider doing the above processing on multiple frames, and then obtain a set of data of parameter candidate values, and select the most stable parameter in this set of data by statistical means.

[0101] In the above method, the image and lidar data are fused using contour information to establish an optimization objective function to achieve the purpose of evaluating the current parameters; the information fusion objective function is solved using perturbation analysis and probability analysis models to evaluate whether the current solution is the optimal solution, so as to evaluate the current parameters; the optimal solution of the objective function is solved using a search algorithm to achieve the goal of calibrating the current parameters. The method of automatically calibrating the parameters between the camera and the lidar online does not require manual marking, complex equipment and calibration processes, and does not require scenes with fixed signs or features. Compared with previous methods, it is more flexible and universal, and the parameters can be updated online. When the current parameters are inaccurate, calibration can be performed, which improves the calibration efficiency. In response to the slow changes in the external posture of a sensor, it consumes very little CPU time, has a fast reaction speed, and significantly improves the accuracy of the perception data. There is no need to know the current calibration external parameters in advance, and an alarm notification and repair processing are issued through real-time evaluation. Compared with directly solving nonlinear problems, it saves more time and computing resources, and greatly improves the efficiency of evaluation and calibration.

[0102] Corresponding to the above method embodiment, this embodiment also provides a parameter evaluation device, such as Figure 4 As shown, the device comprises:

[0103] An acquisition module 41 is used to acquire a first edge map and a second edge map of a target scene; wherein the first edge map is obtained by a camera device; and the second edge map is obtained by a laser radar;

[0104] A determination module 42 is used to determine the correlation between the first edge map and the second edge map corresponding to each set of parameters based on multiple sets of parameters; wherein the multiple sets of parameters include current parameters and multiple sets of perturbation parameters obtained by performing perturbation analysis on the current parameters;

[0105] The evaluation module 43 is used to determine the evaluation result of the current parameters according to the correlation degree corresponding to each group of parameters.

[0106] The present invention provides a parameter evaluation device, which obtains a first edge map and a second edge map of a target scene; wherein the first edge map is obtained by a camera device; the second edge map is obtained by a laser radar; multiple groups of parameters are obtained to determine the correlation between the first edge map and the second edge map corresponding to each group of parameters; wherein the multiple groups of parameters include current parameters, and multiple groups of perturbation parameters obtained by performing perturbation analysis on the current parameters; the parameters are used to indicate the posture relationship between the camera device and the laser radar; and the evaluation result of the current parameter is determined according to the correlation corresponding to each group of parameters. In this method, multiple groups of parameters are obtained through perturbation analysis, and the corresponding correlation is determined based on each group of parameters, so that the accuracy of the current parameter can be determined in real time. Compared with the evaluation and marking method in an offline state, the accuracy of the current parameter can be evaluated in real time, thereby improving the accuracy of the perception data.

[0107] Furthermore, the above-mentioned determination module is also used to: for each set of parameters, project the second edge map to the first edge map through the parameters, remove the points of the second edge map outside the first edge map, and obtain a third edge map corresponding to the second edge map; according to the second edge map and the third edge map corresponding to the second edge map, calculate the correlation between the first edge map and the second edge map corresponding to each set of parameters.

[0108] Furthermore, the above-mentioned second edge map includes multiple images; the above-mentioned determination module is also used to: for each set of parameters, calculate the product value of the depth value of each pixel point in each second edge map and the depth value of the pixel point at the corresponding position in the third edge map corresponding to each second edge map; and determine the sum of the product values ​​corresponding to each second edge map as the correlation degree between the first edge map and the second edge map corresponding to the parameter.

[0109] Furthermore, the above-mentioned correlation degree includes: the correlation degree corresponding to the current parameter and the correlation degree corresponding to the disturbance parameter; the above-mentioned evaluation module is also used to: obtain the target correlation degree from the correlation degree corresponding to the disturbance parameter; wherein the target correlation degree is less than the correlation degree corresponding to the current parameter; calculate the ratio of the number of target correlation degrees to the number of correlation degrees corresponding to the disturbance parameter; and determine the ratio as the evaluation result of the current parameter.

[0110] Furthermore, the above-mentioned acquisition module is also used to: collect the original image of the target scene through a camera device, and convert the original image into a grayscale image; obtain the pixel point with the largest grayscale value in each specified area of ​​the grayscale image in a preset order; determine the edge map of the grayscale image based on the pixel point with the largest grayscale value; perform inverse distance transformation on the edge map to obtain a first edge map.

[0111] Furthermore, the above-mentioned acquisition module is also used to: collect point cloud data of the target scene through a laser radar; filter out noise points in the point cloud data according to the depth value of each position point in the point cloud data to obtain filtered point cloud data; obtain edge points in the filtered point cloud data through the depth continuity between each position point in the filtered point cloud data and the adjacent position points; and determine a second edge map based on the edge points.

[0112] Furthermore, the above-mentioned acquisition module is also used to: for each position point in the point cloud data, calculate the absolute value of the first depth difference between the depth value of the position point and the depth value of the previous position point, and the absolute value of the second depth difference between the depth value of the position point and the depth value of the next position point; if the absolute value of the first depth difference and the absolute value of the second depth difference are both greater than the first specified depth threshold, filter out the position point to obtain filtered point cloud data.

[0113] Furthermore, the above-mentioned acquisition module is also used to: calculate, for each position point in the filtered point cloud data, the absolute value of the third depth difference between the depth value of the position point and the depth value of the previous position point, and the absolute value of the fourth depth difference between the depth value of the position point and the depth value of the next position point; obtain the maximum value of the absolute value of the third depth difference and the absolute value of the fourth depth difference; if the maximum value is greater than the second specified depth threshold, determine the position point as an edge point; and determine a second edge map based on the edge point.

[0114] Furthermore, the above-mentioned evaluation results include evaluation indicators; the above-mentioned device also includes a calibration module, which is used to calculate the accuracy of the current parameters based on the mean and standard deviation of the evaluation indicators corresponding to the preset correct parameters, and the mean and standard deviation of the evaluation indicators corresponding to the preset incorrect parameters; if the accuracy of the current parameters is less than the specified threshold, multiple groups of specified parameters are obtained by performing perturbation analysis of the current parameters with a specified perturbation radius; based on the correlation degree corresponding to the current parameters and the correlation degree corresponding to each group of specified parameters, the current parameters are calibrated.

[0115] Furthermore, the above-mentioned calibration module is also used to: calculate multiple evaluation indicators under the specified parameters for the correlation degree corresponding to each group of specified parameters; calculate the mean and standard deviation of multiple evaluation indicators, and determine the specified parameters that meet the preset mean and preset standard deviation as target parameters; calibrate the target parameters to the current parameters.

[0116] The parameter evaluation device provided in the embodiment of the present invention has the same technical features as the parameter evaluation method provided in the above embodiment, and therefore can also solve the same technical problems and achieve the same technical effects.

[0117] This embodiment also provides an electronic device, including a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the above parameter evaluation method.

[0118] See also Figure 5 As shown, the electronic device includes a processor 100 and a memory 101. The memory 101 stores machine executable instructions that can be executed by the processor 100. The processor 100 executes the machine executable instructions to implement the above parameter evaluation method.

[0119] Further, Figure 5 The electronic device shown further includes a bus 102 and a communication interface 103 , and the processor 100 , the communication interface 103 and the memory 101 are connected via the bus 102 .

[0120] The memory 101 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 103 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used. The bus 102 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0121] The processor 100 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 100. The above processor 100 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in conjunction with the embodiments of the present invention can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module may be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 101, and the processor 100 reads the information in the memory 101 and completes the steps of the method of the above embodiment in combination with its hardware.

[0122] This embodiment further provides a machine-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above parameter evaluation method.

[0123] The parameter evaluation method, device and computer program product of the electronic device provided in the embodiments of the present invention include a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the method described in the previous method embodiments. The specific implementation can be found in the method embodiments, which will not be repeated here.

[0124] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0125] In addition, in the description of the embodiments of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0126] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0127] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0128] Finally, it should be noted that the above embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above embodiments, those skilled in the art should understand that any person skilled in the art can still modify the technical solutions recorded in the above embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A parameter evaluation method, characterized in that: include: Acquire a first edge map and a second edge map of a target scene; wherein the first edge map is obtained by a camera device; the second edge map is obtained by a laser radar; the second edge map includes a plurality of images; For each group of the parameters, projecting the second edge map onto the first edge map by the parameters, removing points of the second edge map outside the first edge map, and obtaining a third edge map corresponding to the second edge map; For each group of the parameters, calculating the product value of the depth value of each pixel in each of the second edge images and the depth value of the pixel at the corresponding position in the third edge image corresponding to each of the second edge images; The sum of the product values ​​corresponding to each of the second edge images is determined as the correlation between the first edge image and the second edge image corresponding to the parameter; the multiple groups of parameters include current parameters and multiple groups of perturbation parameters obtained by performing perturbation analysis on the current parameters; the parameters are used to indicate the posture relationship between the camera device and the laser radar; The evaluation result of the current parameter is determined according to the correlation degree corresponding to each group of the parameters.

2. The method according to claim 1, characterized in that The correlation degree includes: the correlation degree corresponding to the current parameter and the correlation degree corresponding to the disturbance parameter; the step of determining the evaluation result of the current parameter according to the correlation degree corresponding to each group of the parameters includes: Obtaining a target correlation degree from the correlation degrees corresponding to the disturbance parameters; wherein the target correlation degree is less than the correlation degree corresponding to the current parameters; The ratio of the number of the target association degrees to the number of association degrees corresponding to the disturbance parameter is calculated; and the ratio is determined as the evaluation result of the current parameter.

3. The method according to claim 1, characterized in that The step of obtaining a first edge map and a second edge map of a target scene includes: Capturing an original image of the target scene by means of the camera device, and converting the original image into a grayscale image; Obtaining the pixel with the largest grayscale value in each designated area of ​​the grayscale image in a preset order; determining the edge image of the grayscale image based on the pixel with the largest grayscale value; The edge map is subjected to inverse distance transformation processing to obtain a first edge map.

4. The method according to claim 1, characterized in that: The step of obtaining a first edge map and a second edge map of a target scene includes: Collecting point cloud data of the target scene by using the laser radar; According to the depth value of each position point in the point cloud data, the noise points in the point cloud data are filtered out to obtain filtered point cloud data; The edge points in the filtered point cloud data are obtained by comparing the depth continuity between each position point in the filtered point cloud data and the adjacent position points; and the second edge map is determined based on the edge points.

5. The method according to claim 4, characterized in that The step of filtering out noise points in the point cloud data according to the depth value of each position point in the point cloud data to obtain filtered point cloud data comprises: For each position point in the point cloud data, calculate an absolute value of a first depth difference between a depth value of the position point and a depth value of a previous position point, and an absolute value of a second depth difference between the depth value of the position point and a depth value of a subsequent position point; If the absolute value of the first depth difference and the absolute value of the second depth difference are both greater than a first specified depth threshold, the position point is filtered out to obtain filtered point cloud data.

6. The method according to claim 4, characterized in that Obtain edge points in the filtered point cloud data by comparing the depth continuity between each position point in the filtered point cloud data and adjacent position points; The step of determining the second edge map based on the edge points comprises: For each position point in the filtered point cloud data, calculate an absolute value of a third depth difference between the depth value of the position point and the depth value of a previous position point, and an absolute value of a fourth depth difference between the depth value of the position point and the depth value of a subsequent position point; Obtaining a maximum value between an absolute value of the third depth difference and an absolute value of the fourth depth difference; if the maximum value is greater than a second specified depth threshold, determining the position point as the edge point; and determining the second edge map based on the edge point.

7. The method according to claim 1, characterized in that The evaluation result includes an evaluation index; after the step of determining the evaluation result of the current parameter according to the correlation degree corresponding to each group of the parameters, the method further includes: Calculate the accuracy of the current parameters based on the mean and standard deviation of the evaluation indicators corresponding to the preset correct parameters and the mean and standard deviation of the evaluation indicators corresponding to the preset incorrect parameters; If the accuracy of the current parameters is less than a specified threshold, performing a perturbation analysis of a specified perturbation radius on the current parameters to obtain multiple groups of specified parameters; The current parameter is calibrated based on the correlation degree corresponding to the current parameter and the correlation degree corresponding to each group of the specified parameters.

8. The method according to claim 7, characterized in that The step of calibrating the current parameter based on the correlation degree corresponding to the current parameter and the correlation degree corresponding to each group of the specified parameters includes: For each group of the specified parameters, the corresponding correlation degree is calculated for multiple evaluation indicators under the specified parameters; Calculate the mean and standard deviation of the multiple evaluation indicators, determine the specified parameters that meet the preset mean and preset standard deviation as target parameters; and calibrate the target parameters to the current parameters.

9. A parameter evaluation device, characterized in that: include: An acquisition module, used to acquire a first edge map and a second edge map of a target scene; wherein the first edge map is obtained by a camera device; the second edge map is obtained by a laser radar; and the second edge map includes a plurality of images; A determination module, for each set of parameters, projects the second edge map onto the first edge map through the parameters, removes the points of the second edge map outside the first edge map, and obtains a third edge map corresponding to the second edge map; for each set of parameters, calculates the product value of the depth value of each pixel point in each second edge map and the depth value of the pixel point at the corresponding position in the third edge map corresponding to each second edge map; and determines the sum of the product values ​​corresponding to each second edge map as the correlation between the first edge map and the second edge map corresponding to the parameters; wherein the multiple sets of parameters include current parameters, and multiple sets of perturbation parameters obtained by performing perturbation analysis on the current parameters; The evaluation module is used to determine the evaluation result of the current parameter according to the correlation degree corresponding to each group of the parameters.

10. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the parameter evaluation method according to any one of claims 1 to 8.

11. A machine-readable storage medium, characterized in that: The machine-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the parameter evaluation method according to any one of claims 1 to 8.

Images