Point cloud model detection method and device, electronic equipment and readable storage medium

Abstract

The application discloses a point cloud model detection method and device, electronic equipment and a readable storage medium, and belongs to the technical field of point cloud models. The point cloud model detection method comprises the following steps: acquiring a point cloud model, and pre-processing the point cloud model. Fusion parameters of the pre-processed point cloud model are acquired. Chamfer distances of the pre-processed point cloud model are acquired. Local consistency parameters of the pre-processed point cloud model are acquired. A first weight of the fusion parameters, a second weight of the chamfer distances and a third weight of the local consistency parameters are acquired. A detection result is acquired based on the fusion parameters, the chamfer distances, the local consistency parameters, the first weight, the second weight and the third weight, so that the detection of the point cloud model is completed.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of point cloud models, and particularly relates to a point cloud model detection method and device, an electronic device and a readable storage medium. BACKGROUND

[0002] In related technologies, a point cloud model has the problems of sparsity, irregularity and difficulty in obtaining a ground truth. Quality evaluation of the point cloud model is mainly applied to evaluation of a point cloud processing algorithm (point cloud compression, point cloud denoising, etc.), and lacks effective evaluation of the quality of the point cloud model itself.

[0003] The evaluation method in related technologies is generally divided into a global measurement method and a local measurement method.

[0004] The global measurement method mainly has two types of earth mover distance (EMD) and chamfer distance. Although the earth mover distance is relatively accurate, it needs to strictly ensure consistency of the number of compared point clouds, and is difficult to be flexibly applied to actual scenes. The chamfer distance does not need to strictly ensure consistency of the number of compared point clouds, and can be flexibly applied to landing scenes, but it is difficult to ensure accuracy. Since local structure information is lacking, the global method is difficult to effectively measure details of the point cloud model.

[0005] For the local measurement method, a commonly used method measures the point cloud model by means of a local evaluation index, such as whether local normals are consistent, whether point cloud distribution is uniform, etc. Although the local method can effectively measure local structure information, it cannot be directly applied to actual scenes as a unified measurement standard due to the lack of comparison of global information. SUMMARY

[0006] The purpose of the embodiments of the application is to provide a point cloud model detection method and device, an electronic device and a readable storage medium, which can solve the problem of how to more effectively detect the quality of a point cloud model from both global and local aspects, take into account both the accuracy and flexibility of point cloud global measurement, ensure consistency of local geometric structure, and can be effectively applied to actual scenes.

[0007] In a first aspect, the embodiments of the application provide a point cloud model detection method, including: obtaining a point cloud model, and preprocessing the point cloud model. Obtaining a fusion parameter of the preprocessed point cloud model. Obtaining a chamfer distance of the preprocessed point cloud model. Obtaining a local consistency parameter of the preprocessed point cloud model. Obtaining a first weight of the fusion parameter, a second weight of the chamfer distance and a third weight of the local consistency parameter. Based on the fusion parameter, the chamfer distance, the local consistency parameter, the first weight, the second weight and the third weight, obtaining a detection result, and completing detection of the point cloud model.

[0008] In a second aspect, an embodiment of the present application provides a point cloud model detection device, comprising a first acquisition module, a second acquisition module, a third acquisition module, a fourth acquisition module, a fifth acquisition module, and a sixth acquisition module. The first acquisition module is configured to acquire a point cloud model and pre-process the point cloud model. The second acquisition module is configured to acquire fusion parameters of the pre-processed point cloud model. The third acquisition module is configured to acquire chamfer distance of the pre-processed point cloud model. The fourth acquisition module is configured to acquire local consistency parameters of the pre-processed point cloud model. The fifth acquisition module is configured to acquire a first weight of the fusion parameters, a second weight of the chamfer distance, and a third weight of the local consistency parameters. The sixth acquisition module is configured to acquire a detection result based on the fusion parameters, the chamfer distance, the local consistency parameters, the first weight, the second weight, and the third weight, and complete detection of the point cloud model.

[0009] In a third aspect, an embodiment of the present application provides an electronic device, comprising a processor and a memory. The memory stores programs or instructions executable on the processor. When the programs or instructions are executed by the processor, the steps of the point cloud model detection method according to the first aspect are implemented.

[0010] In a fourth aspect, an embodiment of the present application provides a readable storage medium. The readable storage medium stores programs or instructions. When the programs or instructions are executed by a processor, the steps of the point cloud model detection method according to the first aspect are implemented.

[0011] The present application acquires fusion parameters, chamfer distance, local consistency parameters, and weights of the above parameters to acquire a detection result of a point cloud model. The point cloud model is detected from two aspects of global and local, which can take into account the accuracy and flexibility of global point cloud measurement, and can also ensure the consistency of local geometric structure. At the same time, the effectiveness of the detection result is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 Fig. 1 shows a flowchart of a point cloud model detection method according to an embodiment of the present application;

[0013] Figure 2 Fig. 2 shows another flowchart of a point cloud model detection method according to an embodiment of the present application;

[0014] Figure 3 Fig. 3 shows a third flowchart of a point cloud model detection method according to an embodiment of the present application;

[0015] Figure 4 Fig. 4 shows a fourth flowchart of a point cloud model detection method according to an embodiment of the present application;

[0016] Figure 5 Fig. 5 shows a fifth flowchart of a point cloud model detection method according to an embodiment of the present application;

[0017] Figure 6 Fig. 6 shows a flowchart of a method for detecting a point cloud model according to an embodiment of the present application;

[0018] Figure 7 Fig. 7 shows a flowchart of a method for detecting a point cloud model according to an embodiment of the present application;

[0019] Figure 8 Fig. 8 shows a flowchart of a method for detecting a point cloud model according to an embodiment of the present application;

[0020] Figure 9 Fig. 9 shows a flowchart of a method for detecting a point cloud model according to an embodiment of the present application;

[0021] Figure 10 Fig. 10 shows a flowchart of a method for detecting a point cloud model according to an embodiment of the present application;

[0022] Figure 11 Fig. 11 shows a flowchart of a method for detecting a point cloud model according to an embodiment of the present application;

[0023] Figure 12 Fig. 12 shows a schematic diagram of a method for detecting a point cloud model according to an embodiment of the present application;

[0024] Figure 13 Fig. 13 shows a structural block diagram of a device for detecting a point cloud model according to an embodiment of the present application;

[0025] Figure 14 Fig. 14 shows a structural block diagram of an electronic device according to an embodiment of the present application;

[0026] Figure 15 Fig. 15 shows a hardware structure of an electronic device according to an embodiment of the present application.

[0027] wherein, Figures 13 to 15 The correspondence between the reference signs and the component names in the accompanying drawings is as follows:

[0028] 100: detection device of cloud model, 110: first acquisition module; 120: second acquisition module; 130: third acquisition module; 140: fourth acquisition module; 150: fifth acquisition module; 160: sixth acquisition module; 1000: electronic device; 1002: processor; 1004: memory; 1100: electronic device; 1101: radio frequency unit; 1102: network module; 1103: audio output unit; 1104: input unit; 11041: graphics processor; 11042: microphone; 1105: sensor; 1106: display unit; 11061: display panel; 1107: user input unit; 11071: touch panel; 11072: other input device; 1108: interface unit; 1109: memory; 1110: processor. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the present application will be clearly described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art belong to the scope of protection of the present application.

[0030] The terms "first", "second" and the like in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second" and the like are usually a class, not limited to the number of objects, for example, the first object can be one or more. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / ", generally represents a "or" relationship between the front and rear associated objects.

[0031] The technical solutions in the embodiments of the present application will be clearly described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art belong to the scope of protection of the present application. Figures 1 to 15 The point cloud model detection method and device, electronic device and readable storage medium provided by the embodiments of the present application are described in detail below by specific embodiments and application scenarios.

[0032] In the embodiments of the present application, a point cloud model detection method is provided, Figure 1 One of the flow diagrams of the point cloud model detection method provided by the embodiments of the present application is shown, as Figure 1 The point cloud model detection method comprises:

[0033] Step 102, acquiring a point cloud model, and preprocessing the point cloud model.

[0034] Step 104, acquiring fusion parameters of the preprocessed point cloud model.

[0035] In step 106, the chamfer distance of the preprocessed point cloud model is obtained.

[0036] In step 108, the local consistency parameter of the preprocessed point cloud model is obtained.

[0037] In step 110, the first weight of the fusion parameter, the second weight of the chamfer distance, and the third weight of the local consistency parameter are obtained.

[0038] In step 112, the detection result is obtained based on the fusion parameter, the chamfer distance, the local consistency parameter, the first weight, the second weight, and the third weight, and the detection of the point cloud model is completed.

[0039] In the embodiment, the detection result of the point cloud model is obtained by obtaining the fusion parameter, the chamfer distance, the local consistency parameter, and the weights of the above parameters. The point cloud model is detected from two aspects of global and local, which can take into account the accuracy and flexibility of global point cloud measurement, and can also ensure the consistency of local geometric structure. At the same time, the effectiveness of the detection result is ensured. The quality of the point cloud model can be effectively measured through the detection result, so that the point cloud model can be used as a unified measurement index for quality evaluation when the point cloud model is applied in actual scenes.

[0040] In the embodiment, the detection result of the point cloud model is obtained by obtaining the fusion parameter, the chamfer distance, the local consistency parameter, and the weights of the above parameters. The mixed quality evaluation system is used to design from two aspects of global and local, and to cover the characteristics of the point cloud as much as possible. For example, uniform distribution, local consistency, global consistency, etc. The final detection result can comprehensively judge the quality of the point cloud, and provide effective technical support for the selection of the point cloud model in actual scene application.

[0041] In the embodiment, the point cloud model can be an obtained point cloud model or a generated point cloud model. For example, the obtained point cloud model can be a point cloud model obtained by a three-dimensional scanning device, and the generated point cloud model can be a point cloud model generated based on an RGBD (RGB and Depth Map) camera point cloud. Moreover, there is no requirement for the ground truth of the point cloud model, so that the point cloud model detection method has a wider application range.

[0042] In some embodiments of the present application, the point cloud model is preprocessed, specifically including:

[0043] The point cloud model is normalized.

[0044] In the embodiment, the preprocessing operation on the point cloud model can effectively reduce the influence of subsequent operations on the point cloud model, so that the detection result is more accurate.

[0045] Specifically, the obtained original point cloud model is subjected to normalization processing. Through the normalization processing, the influence of the scale and geometric rigid transformation of the point cloud model on the point cloud model can be eliminated, so that the subsequent various parameters obtained by the point cloud model are more accurate.

[0046] In some embodiments of the present application, Figure 2 A flowchart of a second detection method of a point cloud model is shown in FIG. 2B. Figure 2 As shown in FIG. 2B, the fusion parameter of the preprocessed point cloud model is obtained, specifically including:

[0047] In step 202, the accuracy of the preprocessed point cloud model is obtained.

[0048] In step 204, the completeness of the preprocessed point cloud model is obtained.

[0049] In step 206, the harmonic mean of the accuracy and the completeness is obtained to obtain the fusion parameter.

[0050] In the embodiment, the fusion parameter F1 score can include two parts, namely the accuracy Accuracy and the completeness Completeness. Specifically, the accuracy refers to the proportion of the true value three-dimensional points that can be matched within the first preset threshold for each three-dimensional point of the point cloud model to be evaluated. Considering that the point cloud true value itself is not complete, the unobservable part in the true value space needs to be estimated, and the accuracy is ignored when estimating the accuracy. The completeness refers to the proportion of the three-dimensional points of the point cloud model to be evaluated that can be matched within the second preset threshold for each three-dimensional point of the point cloud true value. For example, the value of the first preset threshold can include 5mm, and the value of the second preset threshold can include 5mm.

[0051] It can be understood that the accuracy and the completeness are both important evaluation indicators for the point cloud reconstruction effect, but the above two are a pair of opposite indicators. If the point cloud is filled in the entire space, the completeness reaches 100%, but the accuracy will be greatly reduced. If only a small number of absolutely accurate points are retained, a higher accuracy indicator is obtained, but the completeness is greatly reduced.

[0052] In the embodiment, the accuracy and the completeness are harmonized, and the harmonic mean of the accuracy and the completeness, i.e., the fusion parameter, is used as a detection indicator (evaluation indicator). The accuracy and the completeness of the point cloud model are evaluated through the fusion parameter, specifically:

[0053] Let the precision be p and the completeness be r, then the fusion parameter (that is, the F1 score) is the harmonic mean of the precision and the completeness, that is:

[0054]

[0055] where L F1 represents the fusion parameter p represents the precision, and r represents the completeness.

[0056] In this embodiment, the precision and the completeness are fused to obtain the fusion parameter. Then, the detection result of the point cloud model is obtained through the fusion parameter, which can improve the accuracy and effectiveness of the detection result and effectively evaluate the quality of the point cloud model.

[0057] In some embodiments of the present application, Figure 3 A third flowchart of a point cloud model detection method provided by an embodiment of the present application is shown in FIG. 3. Figure 3 As shown in FIG. 3, the chamfer distance of the preprocessed point cloud model is obtained, specifically including:

[0058] In step 302, the radius of the neighborhood of each point in the preprocessed point cloud model is set to be a first percentage distance of the diagonal length of the preprocessed point cloud model.

[0059] In step 304, a covariance matrix is constructed based on the neighborhood information, and the covariance matrix is eigenvalue decomposed to obtain a first eigenvalue, a second eigenvalue and a third eigenvalue.

[0060] In step 306, a feature indicator function is constructed based on the first eigenvalue, the second eigenvalue and the third eigenvalue, the feature indicator function value of each point is obtained, and normalization is performed.

[0061] In step 308, points with a feature indicator function value greater than a first threshold value are extracted as feature points, and the feature points form a feature point set.

[0062] In step 310, farthest point sampling is performed on the feature point set, and the chamfer distance is obtained based on the sampled point cloud and the sampled ground truth point cloud.

[0063] In this embodiment, for each point in the input point cloud model, the first percentage distance of the model diagonal length is taken as the radius of its neighborhood, for example, the first percentage value includes 5%. Then a covariance matrix is constructed based on the neighborhood information, and the covariance matrix is eigenvalue decomposed to obtain three eigenvalues λ1, λ2 and λ3, wherein λ1 represents the first eigenvalue, λ2 represents the second eigenvalue, and λ3 represents the third eigenvalue, λ1≥λ2≥λ3, and a feature indicator function f is constructed according to the eigenvalue combination:

[0064]

[0065] The feature indicator function value of each point is obtained and normalized to [0, 1].

[0066] In this embodiment, for example, the first threshold value includes 0.8, and the points with the feature indicator function greater than 0.8 can be extracted as feature points to form a feature point set.

[0067] In this embodiment, the chamfer distance is constructed based on the feature points. Unlike directly applying the chamfer distance to the point cloud model, the chamfer distance is applied to the feature point set in the above embodiment. At the same time, in order to ensure the consistency of global metrics, the farthest point sampling is performed on the compared feature point set, so that the global evaluation index is not affected by the point cloud density. Finally, the chamfer distance is calculated between the sampled point cloud and the sampled ground truth point cloud, that is:

[0068]

[0069] L chamfer represents the chamfer distance, S1 represents the point cloud model to be evaluated, S2 represents the ground truth point cloud, and x and y represent the feature points in the feature point set.

[0070] In this embodiment, by constructing the feature indicator function, the feature point set with the feature indicator function value greater than the first threshold value is obtained, the chamfer distance is calculated based on the feature point set, and then the detection result of the point cloud model is obtained through the chamfer distance, which can improve the accuracy and effectiveness of the detection result and effectively evaluate the quality of the point cloud model.

[0071] In this embodiment, the chamfer distance is calculated, a global metric method is used, the consistency of the compared point cloud quantity does not need to be strictly guaranteed, and the method can be flexibly applied to landing scenes to improve the breadth of detection application.

[0072] In some embodiments of the present application, Figure 4 A fourth flowchart of a point cloud model detection method provided by an embodiment of the present application is shown in FIG. 4. Figure 4 As shown in FIG. 4, the local consistency parameter of the preprocessed point cloud model is obtained, which specifically includes:

[0073] In step 402, the density value variance of the points in the preprocessed point cloud model is obtained.

[0074] In step 404, the dot product result average of the points in the preprocessed point cloud model is obtained.

[0075] In step 406, the local consistency parameter is obtained based on the density value variance and the dot product result average.

[0076] In this embodiment, the local consistency parameter is represented as:

[0077] L local = 0.5 x Ldistribution +0.5 x L normal

[0078] wherein, L local denotes the local consistency parameter, L distribution denotes the density value variance, L normal denotes the dot product result average.

[0079] In this embodiment, the local consistency parameter is obtained through the density value variance and the dot product result average, and then the detection result of the point cloud model is obtained through the local consistency parameter, which can improve the accuracy and effectiveness of the detection result and effectively evaluate the quality of the point cloud model.

[0080] This embodiment can effectively measure the local structure information by calculating the local consistency parameter, effectively measure the details of the point cloud model, and improve the precision of the detection result.

[0081] In some embodiments of the present application, Figure 5 Figure 5 shows a flowchart of a point cloud model detection method provided by an embodiment of the present application, as shown in the figure, the density value variance of the points in the preprocessed point cloud model is obtained, specifically including: Figure 5

[0082] Step 502, set the radius of the neighborhood of each point in the preprocessed point cloud model as the second percentage distance of the diagonal length of the preprocessed point cloud model.

[0083] Step 504, obtain the density value of each point, and normalize the density value.

[0084] Step 506, based on the normalized density value of each point, obtain the density value variance of all points.

[0085] In this embodiment, for each point in the input point cloud model, the diagonal length of the model is taken as the radius of its neighborhood, the number of points in its neighborhood is counted as a density value of the point, and then the variance of the density of all points is counted as the measurement information of the uniform distribution. For example, the above density value can be normalized to the interval [0, 1], and then the variance is calculated and represented as L distribution .

[0086] In this embodiment, for example, the value of the second percentage can include 5%.

[0087] This embodiment measures the uniform distribution by obtaining the density value of each point and the density value variance, and then obtains the local consistency parameter through the density value variance, and detects the point cloud model through the local consistency parameter, which can improve the accuracy and effectiveness of the detection result and effectively evaluate the quality of the point cloud model. ​

[0088] In some embodiments of the present application, Figure 6 Fig. 6 shows a flowchart of a sixth method for detecting a point cloud model according to an embodiment of the present application. Figure 6 As shown in Fig. 6, the average value of the dot product results of the points in the preprocessed point cloud model is obtained, specifically including:

[0089] In step 602, the normal of each point in the preprocessed point cloud model is obtained.

[0090] In step 604, the radius of the neighborhood of each point is set as the third percentage distance of the diagonal length of the preprocessed point cloud model.

[0091] In step 606, the normal distribution of each point in its neighborhood is obtained.

[0092] In step 608, for any point, the dot product result of the normal of the point and the normals in its neighborhood is obtained.

[0093] In step 610, the average value of the dot product results of all points is obtained, and the average value of the dot product results is normalized.

[0094] In this embodiment, for the input point cloud model, the normal of each point can be calculated by the principal component analysis method (PCA) first. For each point in the input point cloud model, the third percentage distance of the diagonal length of the model is taken as the radius of its neighborhood to evaluate the distribution of the normals in the neighborhood. Specifically, the dot product of the normal of the point and the normals in its neighborhood is calculated and averaged to serve as a measurement index of whether the local normals are consistent. Since the normal itself only has directional significance, the cross product of the normal can be represented by the angle between the normals. The average value of the dot product results is normalized to the interval [0, 1] and represented as L normal .

[0095] In this embodiment, for example, the value of the third percentage can include 5%.

[0096] This embodiment measures the local normal consistency by obtaining the average value of the dot product results. The subsequent local consistency parameter is obtained by the average value of the dot product results, and the detection of the point cloud model is performed by the local consistency parameter, which can improve the accuracy and effectiveness of the detection result and effectively evaluate the quality of the point cloud model.

[0097] In some embodiments of the present application, Figure 7 Fig. 7 shows a flowchart of a seventh method for detecting a point cloud model according to an embodiment of the present application. Figure 7 As shown in Fig. 7, the first weight of the fusion parameter, the second weight of the chamfer distance, and the third weight of the local consistency parameter are obtained, specifically including:

[0098] Step 702, obtaining a first correlation coefficient between any two parameters of the fusion parameter, the chamfer distance and the local consistency parameter.

[0099] Step 704, based on the first correlation coefficient, obtaining an independent weight of the fusion parameter, an independent weight of the chamfer distance and an independent weight of the local consistency parameter.

[0100] Step 706, normalizing the independent weight of the fusion parameter, the independent weight of the chamfer distance and the independent weight of the local consistency parameter respectively to obtain a first weight of the fusion parameter, a second weight of the chamfer distance and a third weight of the local consistency parameter.

[0101] It can be understood that, since the evaluation of the point cloud quality by the fusion parameter, the chamfer distance and the local consistency parameter has some emphasis, it is necessary to assign a weight to each parameter to obtain the final detection result. The independent weight is determined according to the correlation between the indicators and other indicators. The lower the correlation with other indicators, the higher the weight.

[0102] The number of indicators is determined according to the detection parameters. When the point cloud model is a point cloud based on RGBD reconstruction, the indicator i or the indicator j represents any one of the fusion parameter, the chamfer distance, the local consistency parameter or the re-projection error. When the point cloud model is not a point cloud based on RGBD reconstruction, the indicator i or the indicator j represents any one of the fusion parameter, the chamfer distance or the local consistency parameter. If the indicator i is different from the indicator j, the correlation between the indicator i and the indicator j is represented by a correlation coefficient:

[0103]

[0104] Wherein, r ij represents the correlation coefficient of the indicator i and the indicator j, Cov(i,j) represents the covariance of the indicator i and the indicator j, Var(i) represents the variance of the indicator i, and Var(j) represents the variance of the indicator j.

[0105] In this embodiment, the first correlation coefficient between any two parameters of the fusion parameter, the chamfer distance and the local consistency parameter is obtained by the above formula.

[0106] The independent weight of the indicator j is:

[0107]

[0108] Wherein, C j represents the independent weight of the indicator j, and σ jLet represent the standard deviation of index j, and n represent the number of indices. For example, in this embodiment, when n is 3 and i takes values ​​of 1, 2, and 3, it corresponds to three indices: the fusion parameter, the chamfer distance, and the local consistency parameter. The above formula yields the independence weights for obtaining the fusion parameter, the chamfer distance, and the local consistency parameter, respectively.

[0109] Normalize the independent weights of index j:

[0110]

[0111] Among them, W j This represents the normalized independence weight of index j. When n is 3, and k is 1, 2, or 3 respectively, C... k These correspond to the independent weights of three indicators: fusion parameter, chamfer distance, and local consistency parameter. The first weight of the fusion parameter, the second weight of the chamfer distance, and the third weight of the local consistency parameter are obtained from the above formula.

[0112] In this embodiment, the first weight of the fusion parameter, the second weight of the chamfer distance, and the third weight of the local consistency parameter are obtained respectively. By setting the weights, the point cloud model is detected comprehensively. This can take into account both the accuracy and flexibility of the global measurement of the point cloud, and also ensure the consistency of the local geometric structure, thereby achieving an effective evaluation of the quality of the point cloud model.

[0113] In some embodiments of this application, Figure 8 This illustrates the eighth flowchart of the point cloud model detection method provided in this application embodiment, as shown below. Figure 8 As shown, based on fusion parameters, chamfer distance, local consistency parameters, first weight, second weight, and third weight, the detection results are obtained to complete the detection of the point cloud model, specifically including:

[0114] Step 802: Obtain the first product of the fusion parameters and the first weight.

[0115] Step 804: Obtain the second product of the chamfer distance and the second weight.

[0116] Step 806: Obtain the third product of the local consistency parameter and the third weight.

[0117] Step 808: Obtain the first sum of the first product, the second product, and the third product.

[0118] Step 810: Based on the first result, obtain the detection result and complete the detection of the point cloud model.

[0119] In this embodiment, in order to facilitate the comprehensive evaluation of the point cloud quality by the user, the strategy of independence weight is adopted to weight the adopted indexes, and finally the detection result is obtained. The detection result can be a quantitative score of [0, 1]. The closer the quantitative score is to 1, the better the quality of the point cloud model is.

[0120] Specifically, the quantitative score can be represented as:

[0121] S = W1 x L F1 + W2 x L chamfer + W3 x L local

[0122] Wherein, S represents the quantitative score, W1 represents the first weight, W2 represents the second weight, and W3 represents the third weight.

[0123] The embodiment adopts a mixed quality evaluation system, which is designed from two aspects of global and local, covers the characteristics of point cloud distribution uniformity, local consistency and global consistency, and finally obtains a quantitative score as the detection result. The point cloud quality is comprehensively evaluated through the quantitative score. The more the point cloud model meets the distribution uniformity, local consistency and global consistency, the closer the quantitative score is to 1. Through the detection result, the quality of the point cloud model can be effectively measured, so that when the point cloud model is applied in the actual scene, it can be used as a unified measurement index to effectively evaluate the quality of the point cloud model.

[0124] In some embodiments of the present application, Figure 9 Fig. 9 shows a ninth flowchart of the point cloud model detection method provided by the embodiments of the present application, as shown in the figure, the point cloud model detection method further comprises: Figure 9

[0125] Step 902, based on the pre-processed point cloud model, the depth map information after the re-projection of the pre-processed point cloud model is obtained.

[0126] Step 904, for any pixel in the depth map, the first distance between the depth map information and the original data is obtained.

[0127] Step 906, based on the first distance being less than the second threshold, it is determined that the depth map information is accurate.

[0128] Step 908, the re-projection error of the pre-processed point cloud model is obtained, and the re-projection error is the average accuracy of the depth map.

[0129] ​It can be understood that for the point cloud based on the RGBD reconstruction, the embodiment also sets the re-projection error to measure the quality of the point cloud. The re-projection error is that for the reconstructed point cloud, the depth map information (Depth) after the re-projection of the point cloud model is obtained according to the point position information shot by the camera, and then the obtained depth map is compared with the original data obtained by the depth camera by using the L2 distance (that is, the first distance). For any pixel of each depth map, the L2 distance is calculated, and if the L2 distance is less than a second threshold, it is considered accurate. Finally, the average accuracy L of all images is obtained reproj .

[0130] In the embodiment, for example, the value of the second threshold can be 1 cm.

[0131] In the embodiment, for the case that the pre-processed point cloud model is the reconstructed point cloud, the re-projection error is set to measure the point cloud model, and the detection of the point cloud model is performed through the re-projection error, which can improve the accuracy and effectiveness of the detection result and effectively evaluate the quality of the point cloud model.

[0132] In some embodiments of the present application, Figure 10 Fig. 10 shows a flowchart of a point cloud model detection method provided by an embodiment of the present application, as Figure 10 The point cloud model detection method further includes:

[0133] Step 1002, obtaining a second correlation coefficient of the re-projection error with respect to a fusion parameter, a chamfer distance, and a local consistency parameter;

[0134] Step 1004, obtaining an independence weight of the re-projection error based on the second correlation coefficient;

[0135] Step 1006, performing normalization processing on the independence weight of the re-projection error to obtain a fourth weight of the re-projection error.

[0136] In the embodiment, the second correlation coefficient of the re-projection error with respect to the fusion parameter, the chamfer distance, and the local consistency parameter is obtained according to the following formula:

[0137]

[0138] Then, the independence weight of the re-projection error is obtained according to the following formula:

[0139]

[0140] Wherein, n is 4, i is 1, 2, 3, and 4, respectively corresponding to four indexes, and the four indexes are the fusion parameter, the chamfer distance, the local consistency parameter, and the re-projection error.

[0141] The independence weights of the reprojection error are normalized according to the following formula to obtain the fourth weight of the reprojection error:

[0142]

[0143] Where n takes the value 4, and k takes the values ​​1, 2, 3, and 4 respectively, C k Each of the four indicators has an independent weight: fusion parameter, chamfer distance, local consistency parameter, and reprojection error.

[0144] In this embodiment, a fourth weight of the reprojection error is obtained. By setting the reprojection error weight, more accurate point cloud model detection can be performed for the case where the preprocessed point cloud model is the reconstructed point cloud, thereby achieving an effective evaluation of the quality of the point cloud model.

[0145] In some embodiments of this application, Figure 11 This document illustrates step eleven of a flowchart illustrating the point cloud model detection method provided in an embodiment of this application. Figure 11 As shown, based on fusion parameters, chamfer distance, local consistency parameters, first weight, second weight, and third weight, the detection results are obtained to complete the detection of the point cloud model. This also includes:

[0146] Step 1102: Obtain the fourth product of the reprojection error and the fourth weight;

[0147] Step 1104: Obtain the second sum of the first product, the second product, the third product, and the fourth product;

[0148] Step 1106: Based on the second result, the detection result is obtained, and the detection of the point cloud model is completed.

[0149] In this embodiment, for the case where the preprocessed point cloud model is a reconstructed point cloud, reprojection error is added to measure the point cloud model. An independent weighting strategy is used to weight the adopted indicators, which can improve the accuracy and effectiveness of the detection results and effectively assess the quality of the point cloud model. To facilitate users' comprehensive evaluation of the point cloud quality, the detection result can be a quantized score of [0,1]. The closer the quantized score is to 1, the better the quality of the point cloud model.

[0150] Specifically, the quantified score can be expressed as:

[0151] S = W1 × L F1 +W2×L chamfer +W3×L local +W4×L reproj

[0152] Wherein, S represents a quantification score, W1 represents a first weight, W2 represents a second weight, W3 represents a third weight, and W4 represents a fourth weight.

[0153] The embodiment adopts a mixed quality evaluation system, is designed from two levels of global and local, covers characteristics such as uniform distribution, local consistency and global consistency of point cloud, and finally obtains a detection result which can be a quantification score. The quality of point cloud is comprehensively judged through the quantification score. The more the point cloud model meets the uniform distribution, local consistency and global consistency, the closer the quantification score is to 1. The quality of the point cloud model can be effectively measured through the detection result, so that the point cloud model can be used as a unified measurement index for quality evaluation of the quality of the point cloud model in practical scene application.

[0154] In some embodiments of the application, Figure 12 A scheme diagram of the point cloud model detection method provided by the embodiment of the application is shown, as shown in Figure 12 The point cloud model detection method first inputs a point cloud model 1202. Then, parameters are acquired 1204, specifically including: acquiring fusion parameters, chamfer distance, local consistency parameters, re-projection error, first weight, second weight, third weight and fourth weight. Finally, a quantification score is output 1206.

[0155] Wherein, the fusion parameters L F1 , the chamfer distance L chamfer , the local consistency parameters L local , the re-projection error L reproj , the first weight W1, the second weight W2, the third weight W3 and the fourth weight W4 are acquired through the formulas in the above embodiments, and the quantification score is represented as:

[0156] S = W1 x L F1 + W2 x L chamfer + W3 x L local + W4 x L reproj

[0157] The point cloud model detection method provided by the embodiment of the application can be executed by a point cloud model detection device. In the embodiment of the application, the point cloud model detection device is taken as an example to execute the point cloud model detection method, and the point cloud model detection device provided by the embodiment of the application is described.

[0158] A point cloud model detection device is provided in some embodiments of the application, Figure 13 A structure block diagram of the point cloud model detection device provided by the embodiment of the application is shown, as shown in Figure 13As shown, the point cloud model detection apparatus 100 comprises a first obtaining module 110, a second obtaining module 120, a third obtaining module 130, a fourth obtaining module 140, a fifth obtaining module 150 and a sixth obtaining module 160. The first obtaining module 110 is configured to obtain a point cloud model and pre-process the point cloud model. The second obtaining module 120 is configured to obtain fusion parameters of the pre-processed point cloud model. The third obtaining module 130 is configured to obtain chamfer distance of the pre-processed point cloud model. The fourth obtaining module 140 is configured to obtain local consistency parameters of the pre-processed point cloud model. The fifth obtaining module 150 is configured to obtain a first weight of the fusion parameters, a second weight of the chamfer distance and a third weight of the local consistency parameters. The sixth obtaining module 160 is configured to obtain a detection result based on the fusion parameters, the chamfer distance, the local consistency parameters, the first weight, the second weight and the third weight, and complete detection of the point cloud model.

[0159] In this embodiment, the detection result of the point cloud model is obtained by obtaining the fusion parameters, the chamfer distance, the local consistency parameters and the weights of the above parameters. The point cloud model is detected from both global and local aspects, which can take into account the accuracy and flexibility of global point cloud measurement and ensure the consistency of local geometric structure. At the same time, the effectiveness of the detection result is ensured. The quality of the point cloud model can be effectively measured through the detection result, so that the point cloud model can be used as a unified measurement index for quality evaluation of the point cloud model when the point cloud model is applied in an actual scene.

[0160] The point cloud model detection apparatus in the embodiments of the present application can be an electronic device or a component in an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices other than a terminal. For example, the electronic device can be a mobile phone, a tablet computer, a notebook computer, a palm computer, a vehicle-mounted electronic device, a mobile Internet device (MID), an augmented reality (AR) / virtual reality (VR) device, a robot, a wearable device, an ultra-mobile personal computer (UMPC), a netbook or a personal digital assistant (PDA), etc. The electronic device can also be a server, a network attached storage (NAS), a personal computer (PC), a television (TV), a teller machine or a self-service machine, etc. The embodiments of the present application are not limited in this regard.

[0161] The point cloud model detection device in this application embodiment can be a device with an operating system. This operating system can be Android, iOS, or other possible operating systems; this application embodiment does not specifically limit it.

[0162] The point cloud model detection device provided in this application embodiment can implement all the processes implemented in the above method embodiments, and will not be described again here to avoid repetition.

[0163] Optionally, such as Figure 14 As shown, this application embodiment also provides an electronic device 1000, which includes a processor 1002 and a memory 1004. The memory 1004 stores a program or instructions that can run on the processor 1002. When the program or instructions are executed by the processor 1002, they implement the various steps of the above method embodiments and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0164] It should be noted that the electronic devices in the embodiments of this application include the aforementioned mobile electronic devices and non-mobile electronic devices.

[0165] Figure 15 A schematic diagram of the hardware structure of an electronic device to implement an embodiment of this application.

[0166] The electronic device 1100 includes, but is not limited to, components such as: radio frequency unit 1101, network module 1102, audio output unit 1103, input unit 1104, sensor 1105, display unit 1106, user input unit 1107, interface unit 1108, memory 1109, and processor 1110.

[0167] Those skilled in the art will understand that the electronic device 1100 may also include a power supply (such as a battery) for supplying power to various components. The power supply may be logically connected to the processor 1110 through a power management system, thereby enabling functions such as managing charging, discharging, and power consumption through the power management system. Figure 15 The electronic device structure shown does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown, or combine certain components, or have different component arrangements, which will not be elaborated here.

[0168] The processor 1110 is used to acquire the point cloud model and preprocess the point cloud model.

[0169] Processor 1110 is used to obtain the fusion parameters of the preprocessed point cloud model.

[0170] Processor 1110 is used to obtain the chamfer distance of the preprocessed point cloud model.

[0171] The processor 1110 is configured to acquire the local consistency parameter of the preprocessed point cloud model.

[0172] The processor 1110 is configured to acquire a first weight of the fusion parameter, a second weight of the chamfer distance, and a third weight of the local consistency parameter.

[0173] The processor 1110 is configured to acquire the detection result based on the fusion parameter, the chamfer distance, the local consistency parameter, the first weight, the second weight, and the third weight, to complete the detection on the point cloud model.

[0174] In this embodiment, the detection result of the point cloud model is acquired by acquiring the fusion parameter, the chamfer distance, the local consistency parameter, and the weights of the above parameters. The point cloud model is detected from two aspects of global and local, which can take into account the accuracy and flexibility of the global point cloud measurement, and can also ensure the consistency of the local geometric structure. At the same time, the effectiveness of the detection result is ensured. The quality of the point cloud model can be effectively measured through the detection result, so that the point cloud model can be used as a unified measurement index for quality evaluation when the point cloud model is applied in an actual scene.

[0175] The processor 110 provided in the embodiments of the present application can implement each process of the point cloud model detection method embodiments described above, and achieve the same technical effects. To avoid repetition, details are not described here.

[0176] It should be understood that, in the embodiments of the present application, the input unit 1104 can include a graphics processing unit (GPU) 11041 and a microphone 11042. The graphics processing unit 11041 processes image data of a still picture or a video obtained by an image capture device (such as a camera) in a video capture mode or an image capture mode. The display unit 1106 can include a display panel 11061, which can be configured in the form of a liquid crystal display, an organic light-emitting diode, etc. The user input unit 1107 includes at least one of a touch panel 11071 and other input devices 11072. The touch panel 11071 is also called a touch screen. The touch panel 11071 can include a touch detection device and a touch controller. The other input devices 11072 can include, but are not limited to, a physical keyboard, function keys (such as volume control keys, on-off keys, etc.), trackballs, mice, joysticks, and the like, which are not described here.

[0177] The memory 1109 can be used to store software programs and various data. The memory 1109 can mainly include a first storage area storing programs or instructions and a second storage area storing data, wherein the first storage area can store an operating system, application programs or instructions required by at least one function (such as a sound playing function, an image playing function, etc.), etc. In addition, the memory 1109 can include a volatile memory or a non-volatile memory, or the memory 1109 can include both volatile and non-volatile memories. The non-volatile memory can be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically EPROM (EEPROM), or a flash memory. The volatile memory can be a Random Access Memory (RAM), a Static RAM (SRAM), a Dynamic RAM (DRAM), a Synchronous DRAM (SDRAM), a Double Data Rate SDRAM (DDR SDRAM), an Enhanced SDRAM (ESDRAM), a Synch link DRAM (SLDRAM), and a Direct Rambus RAM (DRRAM). The memory 1109 in the embodiments of the present application includes but is not limited to these and any other suitable types of memories.

[0178] The processor 1110 can include one or more processing units; optionally, the processor 1110 integrates an application processor and a modem processor, wherein the application processor mainly processes operations related to an operating system, a user interface, and an application program, and the modem processor mainly processes wireless communication signals, such as a baseband processor. It can be understood that the above-mentioned modem processor can also not be integrated into the processor 1110.

[0179] The embodiments of the present application also provide a readable storage medium, and the readable storage medium stores programs or instructions, the programs or instructions are executed by the processor to realize each process of the above-mentioned point cloud model detection method embodiment, and the same technical effects can be achieved. To avoid repetition, it will not be repeated here.

[0180] The processor is a processor in the electronic device in the above embodiments. The readable storage medium includes a computer readable storage medium, such as a computer readable only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0181] The chip provided in the embodiments of the present application includes a processor and a communication interface. The communication interface is coupled with the processor. The processor is configured to execute programs or instructions to implement various processes of the detection method of the point cloud model and achieve the same technical effects. To avoid repetition, details are not described herein.

[0182] It should be understood that the chip mentioned in the embodiments of the present application can also be referred to as a system-level chip, a system chip, a chip system, or a system-on-chip, etc.

[0183] The embodiments of the present application provide a computer program product stored in a storage medium. The program product is executed by at least one processor to implement various processes of the detection method of the point cloud model and achieve the same technical effects. To avoid repetition, details are not described herein.

[0184] It should be noted that in this document, the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, articles, or devices including a series of elements not only include those elements, but also include other elements not explicitly listed, or include elements inherent to such processes, methods, articles, or devices. Without more limitations, the element defined by the statement "including a" does not exclude the presence of additional identical elements in the process, method, article, or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to the order of performing the functions as shown or discussed, but can also include performing the functions in a substantially simultaneous manner or in a reverse order, for example, the described method can be performed in an order different from the described order, and various steps can be added, omitted, or combined. In addition, the features described with reference to certain examples can be combined in other examples.

[0185] From the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment method can be realized by means of software and necessary general hardware platform, of course, it can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, or optical disk), and includes a plurality of instructions for making a terminal (which can be a mobile phone, computer, server, or network device, etc.) execute the method of each embodiment of the present application.

[0186] The embodiments of the present application are described above with reference to the accompanying drawings, but the present application is not limited to the above-described specific embodiments, and the above-described specific embodiments are merely illustrative, but not restrictive, and a person of ordinary skill in the art can make many forms under the inspiration of the present application without departing from the purpose of the present application and the scope protected by the claims.

Claims

1. A method for detecting a point cloud model, characterized in that, The method comprises the following steps: obtaining a point cloud model, preprocessing the point cloud model; obtaining a fusion parameter of the preprocessed point cloud model; obtaining a chamfer distance of the preprocessed point cloud model, the fusion parameter and the chamfer distance being used for global measurement of the point cloud model; obtaining a local consistency parameter of the preprocessed point cloud model, the local consistency parameter being used for local measurement of the point cloud model; obtaining a first weight of the fusion parameter, a second weight of the chamfer distance and a third weight of the local consistency parameter; obtaining a detection result based on the fusion parameter, the chamfer distance, the local consistency parameter, the first weight, the second weight and the third weight, and completing detection of the point cloud model.

2. The method of claim 1, wherein, The preprocessing of the point cloud model comprises the following steps: normalizing the point cloud model.

3. The method of claim 1, wherein, The obtaining of the fusion parameter of the preprocessed point cloud model comprises the following steps: obtaining a precision of the preprocessed point cloud model; obtaining a completeness of the preprocessed point cloud model; obtaining a harmonic mean of the precision and the completeness to obtain the fusion parameter.

4. The method of claim 1, wherein, The obtaining of the chamfer distance of the preprocessed point cloud model comprises the following steps: setting a radius of a neighborhood of each point in the preprocessed point cloud model as a first percentage distance of a diagonal length of the preprocessed point cloud model; constructing a covariance matrix based on neighborhood information, performing eigenvalue decomposition on the covariance matrix to obtain a first eigenvalue, a second eigenvalue and a third eigenvalue; constructing a feature indicator function based on the first eigenvalue, the second eigenvalue and the third eigenvalue, obtaining a feature indicator function value of each point, and normalizing the feature indicator function value; extracting points with a feature indicator function value greater than a first threshold value as feature points, the feature points forming a feature point set; performing farthest point sampling on the feature point set, and obtaining the chamfer distance based on the sampled point cloud and a sampled ground truth point cloud.

5. The method of claim 1, wherein, The obtaining of the local consistency parameter of the preprocessed point cloud model comprises the following steps: obtaining a density value variance of points in the preprocessed point cloud model; obtaining a dot product result average value of points in the preprocessed point cloud model based on a dot product result of a normal of any point in the preprocessed point cloud model and normals in a neighborhood of the any point; obtaining the local consistency parameter based on the density value variance and the dot product result average value.

6. The method of claim 5, wherein, The obtaining of the density value variance of points in the preprocessed point cloud model comprises the following steps: setting a radius of a neighborhood of each point in the preprocessed point cloud model as a second percentage distance of a diagonal length of the preprocessed point cloud model; obtaining a density value of each point based on a number of points in a neighborhood of each point, and normalizing the density value; obtaining the density value variance of all points based on the normalized density value of each point.

7. The method of claim 5, wherein, The obtaining of the dot product result average value of points in the preprocessed point cloud model comprises the following steps: obtaining a normal of each point in the preprocessed point cloud model; setting a radius of a neighborhood of each point as a third percentage distance of a diagonal length of the preprocessed point cloud model; obtaining a normal distribution of each point in the neighborhood of the point; For any point, the dot product result of the normal of the arbitrary point and the normal in the neighborhood of the arbitrary point is obtained; An average of the dot product results of all points is obtained, and the average of the dot product results is normalized.

8. The method of claim 1, wherein, The first weight of the fusion parameter, the second weight of the chamfer distance, and the third weight of the local consistency parameter are obtained, and the first weight of the fusion parameter, the second weight of the chamfer distance, and the third weight of the local consistency parameter are obtained. A first correlation coefficient between any two parameters of the fusion parameter, the chamfer distance, and the local consistency parameter is obtained. Based on the first correlation coefficient, an independence weight of the fusion parameter, an independence weight of the chamfer distance, and an independence weight of the local consistency parameter are obtained. The independence weight of the fusion parameter, the independence weight of the chamfer distance, and the independence weight of the local consistency parameter are normalized respectively to obtain the first weight of the fusion parameter, the second weight of the chamfer distance, and the third weight of the local consistency parameter.

9. The method of claim 8, wherein, Based on the fusion parameter, the chamfer distance, the local consistency parameter, the first weight, the second weight, and the third weight, a detection result is obtained, and the detection of the point cloud model is completed, and the detection result is obtained, and the detection of the point cloud model is completed. A first product of the fusion parameter and the first weight is obtained. A second product of the chamfer distance and the second weight is obtained. A third product of the local consistency parameter and the third weight is obtained. A first sum of the first product, the second product, and the third product is obtained. Based on the first sum, a detection result is obtained, and the detection of the point cloud model is completed.

10. The method of claim 1 to 9, wherein, Before the local consistency parameter of the preprocessed point cloud model is obtained, the following steps are further included: Based on the preprocessed point cloud model being a reconstructed point cloud, depth map information after re-projection of the preprocessed point cloud model is obtained. For any pixel in the depth map, a first distance between the depth map information and the original data is obtained. Based on the first distance being less than a second threshold, it is determined that the depth map information is accurate. A re-projection error of the preprocessed point cloud model is obtained, and the re-projection error is an average accuracy of the depth map.

11. The method of claim 10, wherein, Further comprising: A second correlation coefficient of the re-projection error with the fusion parameter, the chamfer distance, and the local consistency parameter is obtained. Based on the second correlation coefficient, an independence weight of the re-projection error is obtained. The independence weight of the re-projection error is normalized to obtain a fourth weight of the re-projection error.

12. The method of claim 11, wherein, Based on the fusion parameter, the chamfer distance, the local consistency parameter, the first weight, the second weight, and the third weight, a detection result is obtained, and the detection of the point cloud model is completed. A fourth product of the re-projection error and the fourth weight is obtained. A second sum of the first product, the second product, the third product, and the fourth product is obtained. Based on the second sum, a detection result is obtained, and the detection of the point cloud model is completed.

13. A device for detecting a point cloud model, characterized in that, The first acquisition module is configured to obtain a point cloud model, and the point cloud model is preprocessed. The second acquisition module is configured to obtain a fusion parameter of a preprocessed point cloud model. ​ The third obtaining module is configured to obtain a chamfer distance of the preprocessed point cloud model, and the fusion parameter and the chamfer distance are used for global measurement of the point cloud model. The fourth obtaining module is configured to obtain a local consistency parameter of the preprocessed point cloud model, and the local consistency parameter is used for local measurement of the point cloud model. The fifth obtaining module is configured to obtain a first weight of the fusion parameter, a second weight of the chamfer distance, and a third weight of the local consistency parameter. The sixth obtaining module is configured to obtain a detection result based on the fusion parameter, the chamfer distance, the local consistency parameter, the first weight, the second weight, and the third weight, and complete detection of the point cloud model.

14. An electronic device, comprising: comprise: a memory having a program or instruction stored thereon; a processor configured to implement the steps of the point cloud model detection method according to any one of claims 1 to 12 when executing the program or instruction.

15. A readable storage medium, having stored thereon a program or instructions, characterized in that, The program or instruction, when executed by the processor, implements the steps of the point cloud model detection method according to any one of claims 1 to 12.

Images