A method, device, electronic device and storage medium for identifying a portal crane

Abstract

The present invention relates to a method, device, electronic device and storage medium for identifying a portal crane, belonging to the technical field of construction machinery. The method for identifying a portal crane includes: determining key points and neighborhood points based on the acquired point cloud data of the portal crane to be identified, constructing a spherical neighborhood centered on the key points based on the established local reference coordinate system, dividing the spherical neighborhood to obtain a plurality of subspaces, and calculating the cosine value of the angle between the neighborhood points and the key points within the subspaces. A histogram is obtained based on the cosine value of the angle, and the SHOT descriptor is obtained by calculating the histogram using the bilinear interpolation method. The SHOT descriptor is encoded using the QB-SHOT method to obtain the QB-SHOT descriptor. The portal crane components corresponding to the key points are determined based on the QB-SHOT descriptor, retaining important feature information and improving the identification accuracy of the portal crane.

Description

Technical Field

[0001] The present invention relates to the technical field of construction machinery, and in particular to a method, device, electronic device and storage medium for identifying a portal crane. Background Art

[0002] With the development of port automation, unmanned operation and intelligence, higher requirements are put forward for the identification and positioning of portal cranes. Traditional two-dimensional image vision methods are easily affected by environmental factors such as light and occlusion in ports, resulting in serious deficiencies in recognition accuracy and robustness. Existing 3D object recognition methods can generally be divided into two categories: one is the recognition method based on global features, and the other is the recognition method based on local features.

[0003] In the three-dimensional object target recognition based on local features, feature description is the encoding of the surface geometric information and spatial distribution of the neighborhood of feature points, which is the core step in the three-dimensional object recognition process. Essentially, a feature descriptor is the encoding of the neighborhood information of feature points.

[0004] The B-SHOT feature descriptor is the binary encoding of the SHOT descriptor. The B-SHOT descriptor mainly divides the specific results according to the proportion of the sum of the values of the members in the same group, but there may be a problem that members with large numerical differences are assigned the same result, resulting in information loss during feature matching. The DB-SHOT adds the numerical comparison between the members of the same group on the basis of the B-SHOT feature descriptor, which increases the discrimination of the members of the same group to a certain extent, but the same result may still appear in two different groups, and there will also be a problem of information loss during feature matching. Summary of the Invention

[0005] In view of this, it is necessary to provide a method, device, electronic device and storage medium for identifying a portal crane to solve the technical problem of information loss when identifying a portal crane.

[0006] To solve the above problems, the present invention provides a method for identifying a portal crane, including:

[0007] Determining key points and neighborhood points based on the obtained point cloud data of the portal crane to be identified;

[0008] Constructing a spherical neighborhood centered on the key point based on the established local reference coordinate system, dividing the spherical neighborhood to obtain a plurality of subspaces, and calculating the cosine value of the angle between the neighborhood points and the key point in the subspace, and obtaining a histogram based on the cosine value;

[0009] The histogram is calculated using the bilinear interpolation method to obtain the SHOT descriptor, and the QB-SHOT method is used to encode the SHOT descriptor to obtain the QB-SHOT descriptor. Based on the QB-SHOT descriptor, the portal crane component corresponding to the key point is determined.

[0010] In a possible implementation, the establishment of the local reference coordinate system includes:

[0011] Obtain the neighborhood points within the neighborhood of the key point, and calculate the centroid of the neighborhood points based on the neighborhood points;

[0012] Obtain the key point, and determine the distance weight based on the key point and the neighborhood points;

[0013] Construct a covariance matrix based on the centroid of the neighborhood points, the distance weight, the key point and the neighborhood points, perform eigenvalue decomposition on the covariance matrix to obtain the eigenvectors and the eigenvalues corresponding to the eigenvectors;

[0014] Determine the x-axis, y-axis, and z-axis of the local reference coordinate system based on the eigenvectors and the eigenvalues.

[0015] In a possible implementation, the construction of the spherical neighborhood centered on the key point based on the constructed local reference coordinate system includes:

[0016] Determine the neighborhood radius of the neighborhood points and the key point;

[0017] Construct a spherical neighborhood centered on the key point and with the neighborhood radius as the radius based on the local reference coordinate system.

[0018] In a possible implementation, the division of the spherical neighborhood to obtain multiple subspaces includes:

[0019] Divide the spherical neighborhood along the radial, latitude, and longitude directions to obtain multiple subspaces.

[0020] In a possible implementation, the calculation of the cosine value of the angle between the neighborhood points and the key point within the subspace and obtaining the histogram based on the cosine value of the angle includes:

[0021] Calculate the normal of the key point within the subspace;

[0022] Calculate the normal of the neighborhood points within the subspace;

[0023] Determine the cosine value of the angle between the neighborhood points and the key point within the subspace based on the normal of the key point and the normal of the neighborhood points;

[0024] Equalize the cosine values of the angles between the neighborhood points and the key points to obtain a histogram, where the histogram uses the cosine value of the angle as the abscissa and the quantity as the ordinate.

[0025] In a possible implementation, the method of using bilinear interpolation to calculate the histogram to obtain the SHOT descriptor includes:

[0026] Obtain the first descriptor of the key point based on the histogram,

[0027] Use the bilinear interpolation method to calculate the first descriptor of the key point to obtain the SHOT descriptor of the key point, where the SHOT descriptor contains the statistical information of the cosine value of the angle between the normal of the neighborhood point and the normal of the key point.

[0028] In a possible implementation, the method of using the QB-SHOT method to encode the SHOT descriptor to obtain the QB-SHOT descriptor includes:

[0029] Initialize the SHOT descriptor to obtain the value range of the SHOT descriptor;

[0030] Preset the first interval threshold, the second interval threshold, the third interval threshold, and the fourth interval threshold, where the second interval threshold is greater than the first interval threshold, the third interval threshold is greater than the second interval threshold, and the fourth interval threshold is greater than the third interval threshold;

[0031] Divide the value range of the SHOT descriptor based on the first interval threshold, the second interval threshold, the third interval threshold, and the fourth interval threshold to obtain the first value range, the second value range, the third value range, and the fourth value range;

[0032] Use the QB-SHOT method to perform binary conversion with two coding bits on the first value range, the second value range, the third value range, and the fourth value range to obtain the first code and the second code, and synthesize the first code and the second code to obtain the QB-SHOT descriptor. After performing binary conversion with two coding bits on the first value range, the first code and the second code are 0. After performing binary conversion with two coding bits on the second value range, the first code is 0 and the second code is 1. After performing binary conversion with two coding bits on the third value range, the first code is 1 and the second code is 0. After performing binary conversion with two coding bits on the fourth value range, the first code and the second code are 1.

[0033] On the other hand, the present invention also provides a portal crane recognition device, including:

[0034] The key point determination module is used to determine key points and neighborhood points based on the acquired point cloud data of the portal crane to be recognized;

[0035] The histogram acquisition module is used to construct a spherical neighborhood centered on the key point based on the constructed local reference coordinate system, divide the spherical neighborhood to obtain multiple subspaces, calculate the cosine value of the angle between the neighborhood points and the key point within the subspace, and obtain a histogram based on the cosine value of the angle;

[0036] The portal crane recognition module is used to calculate the histogram by using the bilinear interpolation method to obtain the SHOT descriptor, encode the SHOT descriptor by using the QB-SHOT method to obtain the QB-SHOT descriptor, and determine the portal crane component corresponding to the key point based on the QB-SHOT descriptor.

[0037] On the other hand, the present invention also provides an electronic device, including: a processor and a memory;

[0038] The memory stores a computer-readable program executable by the processor;

[0039] When the processor executes the computer-readable program, the steps in the above-mentioned portal crane recognition method are implemented.

[0040] On the other hand, the present invention also provides a computer-readable storage medium, and the computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps in the above-mentioned portal crane recognition method.

[0041] The beneficial effects of the present invention are as follows: a spherical neighborhood centered on the key point is constructed based on the local reference coordinate system, the key point and the neighborhood points are projected into the local reference coordinate system to eliminate the influence of rotation and translation, so that the descriptor of the key point has rotation and translation invariance, the spherical neighborhood is divided to obtain multiple subspaces, and the feature information in the local area can be statistically analyzed more finely, improving the expression ability of the descriptor. Through the histogram, the distribution of the normal vectors of the neighborhood points in the local area is statistically analyzed, and the geometric features of the surface of the portal crane component can be effectively described. The bilinear interpolation method is used to calculate the histogram to obtain the SHOT descriptor, and the QB-SHOT method is used to encode the SHOT descriptor, which reduces the calculation amount while retaining important feature information and improves the recognition accuracy of the portal crane. Description of the Drawings

[0042] Figure 1 It is a flowchart of an embodiment of the portal crane recognition method provided by the present invention;

[0043] Figure 2Schematic diagram of the model point cloud and the scene point cloud of the portal crane recognition method provided by the present invention;

[0044] Figure 3 Schematic diagram of the recognized portal crane of the portal crane recognition method provided by the present invention;

[0045] Figure 4 Schematic structural diagram of an embodiment of the portal crane recognition device provided by the present invention;

[0046] Figure 5 Schematic structural diagram of an embodiment of the electronic device provided by the present invention. Detailed implementation manners

[0047] The preferred embodiments of the present invention will be specifically described below with reference to the accompanying drawings. The accompanying drawings form a part of this application and are used together with the embodiments of the present invention to explain the principles of the present invention, rather than to limit the scope of the present invention.

[0048] The present invention discloses a portal crane recognition method, device, electronic device and storage medium, which can be used in a computer. The method, device or computer-readable storage medium involved in the present invention can either be integrated with the above-mentioned device or be relatively independent.

[0049] A specific embodiment of the present invention discloses a portal crane recognition method, which can be executed by a computer, specifically by one or more processors of the computer. Figure 1 is the flowchart of the portal crane recognition method provided by the embodiment of the present invention. Please refer to Figure 1 , the portal crane recognition method includes:

[0050] S101. Determine key points and neighborhood points based on the acquired point cloud data of the portal crane to be recognized;

[0051] S102. Construct a spherical neighborhood centered on the key point based on the constructed local reference coordinate system, divide the spherical neighborhood to obtain a plurality of subspaces, calculate the cosine value of the angle between the neighborhood points and the key point within the subspace, and obtain a histogram based on the cosine value of the angle;

[0052] S103. Calculate the SHOT descriptor by using the bilinear interpolation method for the histogram, encode the SHOT descriptor by using the QB-SHOT method to obtain the QB-SHOT descriptor, and determine the portal crane components corresponding to the key points based on the QB-SHOT descriptor.

[0053] Among them, key points, also known as feature points, are a set of points with stability and distinctiveness obtained from 3D point cloud data by defining detection criteria. The number of key points is much less than the data volume of the original point cloud or image. Combined with local feature descriptors, they form key point descriptors, which are often used to form the representation of the original data, and maintain representativeness and descriptiveness, thus accelerating the subsequent processing speed of data such as recognition and tracking. The QB-SHOT method is used to perform binary encoding on the SHOT descriptor, compensating for the information loss in the binary encoding process of the SHOT descriptor. After determining the portal crane components corresponding to the key points based on the QB-SHOT descriptor, other key points are selected from the point cloud data of the portal crane to be recognized and matched for recognition, and all portal crane components can be recognized.

[0054] Compared with the prior art, the portal crane recognition method provided in this embodiment determines key points and neighborhood points based on the acquired point cloud data of the portal crane to be recognized, constructs a spherical neighborhood centered on the key points based on the constructed local reference coordinate system, divides the spherical neighborhood to obtain multiple subspaces, eliminates the influence of rotation and translation, and makes the descriptor have rotation and translation invariance. Dividing the spherical neighborhood into multiple subspaces can more finely count the feature information in the local area and improve the expression ability of the descriptor. Calculate the cosine value of the angle between the neighborhood points and the key points in the subspace, obtain a histogram based on the cosine value, and through the histogram, count the distribution of the normal vectors of the neighborhood points in the local area, which can effectively describe the geometric features of the surface of the portal crane components. The four-linear interpolation method is used to calculate the histogram to obtain the SHOT descriptor, and the calculation accuracy and stability of the descriptor are improved through four-linear interpolation, thereby improving the accuracy of subsequent point cloud processing. The QB-SHOT method is used to encode the SHOT descriptor to obtain the QB-SHOT descriptor, and the portal crane components corresponding to the key points are determined based on the QB-SHOT descriptor, reducing the calculation amount while retaining important feature information and improving the accuracy of feature matching.

[0055] In some embodiments, in step S101, the lidar point cloud technology emits laser beams and receives reflected signals to obtain the three-dimensional spatial information of the portal crane to be recognized, forming point cloud data, so that the collected data is not affected by factors such as light, weather, and occlusion. The model point cloud and the scene point cloud of the portal crane to be recognized are obtained from the point cloud data. For the schematic diagrams of the model point cloud and the scene point cloud, please refer to Figure 2 , such as Figure 2 shown. Figure 2 The left side is the scene point cloud. Figure 2The right side shows the model point cloud. After the scene point cloud is obtained by laser radar scanning, the scene point cloud is processed and useless points in the model point cloud are removed to obtain the model point cloud. A key point P is selected as the reference point in the model point cloud of the portal crane to be identified, and all the neighboring points in the neighborhood of the key point P are determined. .

[0056] In some embodiments, in step S102, a spherical neighborhood centered on the key point is constructed based on the established local reference coordinate system. First, the local reference coordinate system is established. The process is as follows: determine the key point And the neighboring points: , , , The position coordinates of the neighborhood points are used to calculate the centroid of the neighborhood points. The calculation formula for the centroid of the neighborhood points is:

[0057] ,

[0058] The distance weight is determined based on the key points and neighborhood points. The distance weight is calculated as follows:

[0059] ,

[0060] in, is the distance weight;

[0061] The covariance matrix is constructed based on the centroid, distance weight, key points and neighborhood points of the neighborhood points. The calculation formula of the covariance matrix is:

[0062] ,

[0063] in, is the covariance matrix, is the neighborhood radius, For all points in the neighborhood, is the centroid of all neighborhood points, is the distance from the neighborhood point to the centroid, is the transpose of the matrix;

[0064] Perform eigenvalue decomposition on the covariance matrix to obtain the eigenvector , , , and the eigenvalues corresponding to the eigenvectors , , , , based on the eigenvectors and eigenvalues, determine the x-axis, y-axis, and z-axis of the local reference coordinate system, that is, select the two eigenvectors with the largest eigenvalues and Calculate the eigenvector using the z-axis and x-axis of the local reference coordinate system and Take the cross product as the y-axis of the local reference coordinate system. After obtaining the x-axis, y-axis, and z-axis of the reference coordinate system, the establishment of the local reference coordinate system is initially completed;

[0065] Secondly, construct a spherical neighborhood centered on the key point based on the local reference coordinate system, that is, project the key point P and the neighborhood points into the local reference coordinate system (LRF), determine the neighborhood radius r of the neighborhood points and the key point, and construct a spherical neighborhood space centered on the key point with a radius of r based on the local reference coordinate system. The local reference coordinate system (LRF) increases the uniqueness of the descriptor, converts the key point from the local surface to the local reference coordinate system, and makes the descriptor have rotational and translational invariance;

[0066] Divide the spherical neighborhood in the radial, latitude, and longitude directions, dividing them into 2, 2, and 8 equal parts respectively to obtain 32 subspaces;

[0067] Calculate the cosine value of the angle between the neighborhood points and the key point in the subspace, and obtain a histogram based on the cosine value of the angle. First, for each subspace, calculate the normal vector of the key point in the subspace and the normal vector of the neighborhood points in the subspace, and determine the cosine value of the angle between the neighborhood points and the key point in the subspace based on the normal vector of the key point and the normal vector of the neighborhood points

[0068] .

[0069] Among them, is the normal vector of the key point, is the normal vector of the neighborhood points in the subspace, is the z-axis of the local reference coordinate system;

[0070] Secondly, equally divide the cosine value of the angle between the neighborhood points and the key point to obtain a histogram. Divide the cosine value of the angle into 11 spaces equally, the neighborhood points are assigned to the corresponding intervals, and the number of neighborhood points in each interval is counted to form an 11-dimensional histogram with as the abscissa and the number as the ordinate.

[0071] In some embodiments, in step S103, the histogram is calculated using the bilinear interpolation method to obtain the SHOT descriptor. First, the histograms of 32 subspaces are statistically calculated and combined to obtain the first descriptor of the key point, that is, a 352-dimensional SHOT descriptor of the key point can be obtained. Secondly, the bilinear interpolation method is used to calculate the first descriptor of the key point to obtain the SHOT descriptor of the key point. The specific steps are as follows: for each key point, four directions around the key point are determined according to the key point normal, and then the gradient histograms in the four directions are calculated. Next, the bilinear interpolation method is used to calculate the SHOT descriptor of this point, that is, the gradient histograms in the four directions are used as the values of the interpolation points, and after calculating the distances and weights of the key point in the four directions, the SHOT descriptor of the key point is calculated. Among them, the SHOT descriptor contains the statistical information of the cosine value of the angle between the neighborhood point normal and the key point normal; the SHOT descriptor takes into account the normal vector and curvature information of the point cloud data and is weighted in the calculation. At the same time, a local coordinate system is established during the process of obtaining the SHOT descriptor, making the SHOT descriptor more robust. Therefore, the SHOT feature descriptor has better robustness to noise and local shape changes in the point cloud;

[0072] During the calculation process of the SHOT descriptor, for the neighborhood space of a certain key point, if its subspace does not contain neighborhood points, the corresponding histogram statistical value will be recorded as 0. And the point cloud data is usually distributed in the form of surface points, and there will be cases where the number of neighborhood points in the neighborhood subspace of some key points is 0, resulting in some elements of the SHOT feature vector being 0. And the non-zero elements in the SHOT feature vector indicate the existence of neighborhood points at the corresponding positions. Based on the non-zero terms and distribution intervals of the SHOT feature vector, the QB-SHOT method is used to encode the SHOT descriptor to obtain the QB-SHOT descriptor. The SHOT descriptor is regarded as a group, and the value range of the non-zero terms is divided. According to the interval into which the value of the SHOT descriptor falls, it is encoded as , and iterated in turn. Finally, the descriptor is encoded as . Its QB-SHOT is a SHOT feature descriptor with interval double coding bits, that is, a new coding rule is used to perform binary conversion with double coding bits on the SHOT descriptor. The specific process is as follows: initialize the SHOT descriptor to obtain the value range of the SHOT descriptor , preset the first interval threshold, the second interval threshold, the third interval threshold, and the fourth interval threshold, where the second interval threshold is greater than the first interval threshold, the third interval threshold is greater than the second interval threshold, the fourth interval threshold is greater than the third interval threshold, the first interval threshold is 0, the second interval threshold is greater than 0 and less than or equal to 0.34, the third interval threshold is greater than 0.34 and less than or equal to 0.67, the fourth interval threshold is greater than 0.67 and less than or equal to 1. Based on the first interval threshold, the second interval threshold, the third interval threshold, and the fourth interval threshold, divide the value range of the SHOT descriptor to obtain the first value range, the second value range, the third value range, and the fourth value range, that is, the value of the SHOT descriptor is divided into the interval 、 , , . Among them, the QB-SHOT method is used to perform binary conversion with two coding bits on the first value range, the second value range, the third value range, and the fourth value range to obtain the first code and the second code . After performing binary conversion with two coding bits on the first value range, the values of the first code and the second code are 0. After performing binary conversion with two coding bits on the second value range, the value of the first code is 0, and the value of the second code is 1. After performing binary conversion with two coding bits on the third value range, the value of the first code is 1, and the value of the second code is 0. After performing binary conversion with two coding bits on the fourth value range, the values of the first code and the second code are 1;

[0073] The process of performing binary conversion with two coding bits is as follows:

[0074] When , it means that the number of neighborhood points in the interval is 0, then ;

[0075] When and , then 、 ;

[0076] When and , then 、 ;

[0077] When and , then 、 ;

[0078] The first encoding and the second encoding are synthesized to obtain a SHOT feature descriptor based on interval double-encoded bits, that is, a QB-SHOT descriptor is obtained. Through this encoding method, the loss of information is compensated to a certain extent. For example, for two groups {0.21, 0.37, 0, 0} and {0.98, 0.82, 0, 0}, both will be encoded as {1, 1, 0, 0, 0, 0, 0, 0} using the DB-SHOT rule, while using QB-SHOT, the former group will be encoded as {0, 1, 1, 0, 0, 0, 0, 0} and the latter group will be encoded as {1, 1, 1, 1, 0, 0, 0, 0}, increasing the distinguishability of the encoding results, reducing the probability of incorrect matching, and simplifying the rules of binary encoding to a certain extent.

[0079] Based on the QB-SHOT descriptor, the portal crane component corresponding to the key point is determined. The QB-SHOT descriptor of the key point is matched with the portal crane component descriptors stored in the database to find the portal crane component descriptor that is most similar to the QB-SHOT descriptor of the key point, and then the portal crane component to which the key point P belongs can be determined. Other key points and neighborhood points are selected from the point cloud data of the portal crane to be recognized for encoding and matching, and finally all portal crane components are recognized. For the schematic diagram of the recognized portal crane, please refer to Figure 3 , such as Figure 3 shown, the slashes in the image represent incorrect matches, and the more slashes there are, the lower the recognition rate.

[0080] To better implement the portal crane recognition method in the embodiments of the present invention, correspondingly, as Figure 4 shown, the embodiments of the present invention also provide a portal crane recognition device. The portal crane recognition device 400 includes:

[0081] A key point determination module 401, configured to determine key points and neighborhood points based on the acquired point cloud data of the portal crane to be recognized;

[0082] A histogram acquisition module 402, configured to construct a spherical neighborhood centered on the key point based on the constructed local reference coordinate system, divide the spherical neighborhood to obtain a plurality of subspaces, calculate the cosine value of the angle between the neighborhood points and the key point within the subspace, and obtain a histogram based on the cosine value;

[0083] A portal crane recognition module 403, configured to calculate the SHOT descriptor by using the bilinear interpolation method, encode the SHOT descriptor by using the QB-SHOT method to obtain the QB-SHOT descriptor, and determine the portal crane component corresponding to the key point based on the QB-SHOT descriptor.

[0084] Such asFigure 5 As shown, based on the portal crane recognition method, the present invention also correspondingly provides an electronic device 500, which can be a computing device such as a mobile terminal, a desktop computer, a notebook, a palm computer, and a server. The electronic device 500 includes a processor 501, a memory 502, and a display 503. Figure 5 Only some components of the electronic device 500 are shown, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be alternatively implemented.

[0085] In some embodiments, the memory 502 can be an internal storage unit of the electronic device 500, such as the hard disk or memory of the electronic device 500. In some other embodiments, the memory 502 can also be an external storage device of the electronic device 500, such as a plug-in hard disk equipped on the electronic device 500, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the memory 502 can also include both the internal storage unit and the external storage device of the electronic device 500. The memory 502 is used to store the application software installed on the electronic device 500 and various types of data, such as the program code installed on the electronic device 500. The memory 502 can also be used to temporarily store the data that has been output or will be output. In one embodiment, a portal crane recognition program is stored on the memory 502, and the portal crane recognition program can be executed by the processor 501, thereby implementing the portal crane recognition method of various embodiments of the present invention.

[0086] In some embodiments, the processor 501 can be a central processing unit (CPU), a microprocessor, or other data processing chips, and is used to run the program code stored in the memory 502 or process data, such as the portal crane recognition method, etc.

[0087] In some embodiments, the display 503 can be an LED display, a liquid crystal display, a touch liquid crystal display, and an OLED (Organic Light-Emitting Diode) toucher, etc. The display 503 is used to display the recognition information of the portal crane recognition program and to display a visual user interface. The components 501 - 503 of the electronic device 500 communicate with each other through a system bus.

[0088] In some embodiments, when the processor 501 executes the portal crane recognition program in the memory 502, each step in the portal crane recognition method described in the above embodiments is implemented. Since the portal crane recognition method has been described in detail above, it will not be elaborated here.

[0089] Accordingly, an embodiment of the present application further provides a computer-readable storage medium, which is used to store computer-readable programs or instructions. When the programs or instructions are executed by a processor, the steps or functions in the portal crane recognition method provided by the above various method embodiments can be implemented.

[0090] In summary, for the portal crane recognition method, device, electronic device, and storage medium provided by the present invention, key points and neighborhood points are determined based on the acquired point cloud data of the portal crane to be recognized. A spherical neighborhood centered on the key points is constructed based on the established local reference coordinate system. The spherical neighborhood is divided to obtain multiple subspaces, and the cosine value of the angle between the neighborhood points and the key points within the subspaces is calculated. A histogram is obtained based on the cosine value of the angle. The histogram is calculated using the bilinear interpolation method to obtain the SHOT descriptor. The QB-SHOT method is used to encode the SHOT descriptor to obtain the QB-SHOT descriptor. The portal crane components corresponding to the key points are determined based on the QB-SHOT descriptor, retaining important feature information and improving the recognition accuracy of the portal crane.

[0091] Those skilled in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium is a magnetic disk, an optical disk, a read-only memory, or a random access memory, etc.

[0092] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention.

Claims

1. A method for identifying a portal crane, characterized in that Including: Determining key points and neighborhood points based on the acquired point cloud data of the portal crane to be recognized; Constructing a spherical neighborhood centered on the key point based on the established local reference coordinate system, dividing the spherical neighborhood to obtain multiple subspaces, and calculating the cosine value of the angle between the neighborhood points and the key point within the subspace, and obtaining a histogram based on the cosine value; Calculating the SHOT descriptor by using the bilinear interpolation method for the histogram, encoding the SHOT descriptor by using the QB-SHOT method to obtain the QB-SHOT descriptor, and determining the portal crane component corresponding to the key point based on the QB-SHOT descriptor, wherein encoding the SHOT descriptor by using the QB-SHOT method to obtain the QB-SHOT descriptor includes: Initializing the SHOT descriptor to obtain the value range of the SHOT descriptor; Presetting a first interval threshold, a second interval threshold, a third interval threshold, and a fourth interval threshold, wherein the second interval threshold is greater than the first interval threshold, the third interval threshold is greater than the second interval threshold, and the fourth interval threshold is greater than the third interval threshold; Dividing the value range of the SHOT descriptor based on the first interval threshold, the second interval threshold, the third interval threshold, and the fourth interval threshold to obtain a first value range, a second value range, a third value range, and a fourth value range; Performing binary conversion with double coding bits on the first value range, the second value range, the third value range, and the fourth value range by using the QB-SHOT method to obtain a first code and a second code, and synthesizing the first code and the second code to obtain the QB-SHOT descriptor, wherein after performing binary conversion with double coding bits on the first value range, the first code and the second code are 0, after performing binary conversion with double coding bits on the second value range, the first code is 0 and the second code is 1, after performing binary conversion with double coding bits on the third value range, the first code is 1 and the second code is 0, and after performing binary conversion with double coding bits on the fourth value range, the first code and the second code are 1.

2. The portal crane recognition method according to claim 1, characterized in that, The established local reference coordinate system includes: Obtaining the neighborhood points within the neighborhood of the key point and calculating the centroid of the neighborhood points based on the neighborhood points; Obtaining the key point and determining the distance weight based on the key point and the neighborhood points; Constructing a covariance matrix based on the centroid of the neighborhood points, the distance weight, the key point, and the neighborhood points, and performing eigenvalue decomposition on the covariance matrix to obtain eigenvectors and eigenvalues corresponding to the eigenvectors; Determining the x-axis, y-axis, and z-axis of the local reference coordinate system based on the eigenvectors and the eigenvalues; 3. The portal crane recognition method according to claim 2, wherein The constructing a spherical neighborhood centered on the key point based on the established local reference coordinate system includes: Determining the neighborhood radius of the neighborhood points and the key point; Constructing a spherical neighborhood centered on the key point with the neighborhood radius as the radius based on the local reference coordinate system; 4. The portal crane recognition method according to claim 3, characterized in that The dividing the spherical neighborhood to obtain multiple subspaces includes: Divide the spherical neighborhood in the radial, latitude, and longitude directions to obtain multiple subspaces.

5. The method for identifying a portal crane according to claim 4, wherein Calculating the cosine value of the angle between the neighborhood points and the key points in the subspace, and obtaining a histogram based on the cosine value of the angle, includes: Calculating the normal vector of the key points in the subspace; Calculating the normal vector of the neighborhood points in the subspace; Determining the cosine value of the angle between the neighborhood points and the key points in the subspace based on the normal vector of the key points and the normal vector of the neighborhood points; Equally divide the cosine value of the angle between the neighborhood points and the key points to obtain a histogram, where the histogram uses the cosine value of the angle as the abscissa and the quantity as the ordinate.

6. The portal crane recognition method according to claim 5, characterized in that Calculating the SHOT descriptor by using the bilinear interpolation method for the histogram, includes: Obtaining the first descriptor of the key points based on the histogram, Calculating the SHOT descriptor of the key points by using the bilinear interpolation method for the first descriptor of the key points, where the SHOT descriptor contains the statistical information of the cosine value of the angle between the normal vector of the neighborhood points and the normal vector of the key points.

7. A portal crane identification device, characterized in that, Includes: A key point determination module for determining key points and neighborhood points based on the acquired point cloud data of the portal crane to be recognized; A histogram acquisition module for constructing a spherical neighborhood centered on the key points based on the constructed local reference coordinate system, dividing the spherical neighborhood to obtain multiple subspaces, calculating the cosine value of the angle between the neighborhood points and the key points in the subspace, and obtaining a histogram based on the cosine value of the angle; A portal crane recognition module for calculating the SHOT descriptor by using the bilinear interpolation method for the histogram, encoding the SHOT descriptor by using the QB-SHOT method to obtain a QB-SHOT descriptor, and determining the portal crane component corresponding to the key points based on the QB-SHOT descriptor, where encoding the SHOT descriptor by using the QB-SHOT method to obtain a QB-SHOT descriptor includes: Initializing the SHOT descriptor to obtain the value range of the SHOT descriptor; Presetting a first interval threshold, a second interval threshold, a third interval threshold, and a fourth interval threshold, where the second interval threshold is greater than the first interval threshold, the third interval threshold is greater than the second interval threshold, and the fourth interval threshold is greater than the third interval threshold; Dividing the value range of the SHOT descriptor based on the first interval threshold, the second interval threshold, the third interval threshold, and the fourth interval threshold to obtain a first value range, a second value range, a third value range, and a fourth value range; The QB-SHOT method is used to perform binary conversion with two coding bits on the first value range, the second value range, the third value range, and the fourth value range to obtain a first code and a second code, and the first code and the second code are combined to obtain a QB-SHOT descriptor. Among them, after performing binary conversion with two coding bits on the first value range, both the first code and the second code are 0; after performing binary conversion with two coding bits on the second value range, the first code is 0 and the second code is 1; after performing binary conversion with two coding bits on the third value range, the first code is 1 and the second code is 0; after performing binary conversion with two coding bits on the fourth value range, both the first code and the second code are 1.

8. An electronic device, characterized in that, Comprising a memory and a processor; A computer-readable program executable by the processor is stored on the memory; When the processor executes the computer-readable program, the steps in the portal crane recognition method according to any one of claims 1-6 are implemented.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps in the portal crane recognition method according to any one of claims 1-6.

Images