Point cloud registration method, device, equipment, medium and program product
Through the use of genetic algorithms in Dianyun Matchmaking Standard, obtaining rough standard parameters and combining the ICP algorithm for fine configuration, the ICP algorithm can easily fall into the problem of local optimal solution when dealing with large position errors. Accuracy and robustness.
Patent Information
- Application Number
- CN202411391297.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-08
- Publication Date
- 2025-01-24
AI Technical Summary
In the prior art, ICP algorithms are prone to fall into local optimal solutions when processing large pose errors, and cannot obtain ideal point cloud registration results.
Use the genetic algorithm to obtain the rough allocation parameters between the initial point cloud and the model point cloud of the target workpiece, avoid local optimal solutions through global search capabilities, and use the ICP algorithm on the basis of the rough standard parameters.
Effectively avoid local optimal solutions, improve the accuracy and robustness of the cloud allocation, and ensure the global optimal of allocation.
Smart Images
Figure CN119359773A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of three-dimensional point cloud processing, and in particular to a point cloud registration method, device, equipment, medium and program product. Background Art
[0002] 3D point cloud is the surface data of an object collected by a laser scanner or binocular vision system, which can be widely used in reverse engineering and 3D reconstruction. In practical applications, due to the different perspectives or devices of the point cloud source, the point clouds of the same object have different postures, so it is necessary to align the point clouds of different perspectives or devices through point cloud registration technology to form a complete 3D model.
[0003] At present, the iterative closest point (ICP) algorithm is usually used for point cloud registration. However, the ICP algorithm is prone to fall into a local optimal solution when dealing with large pose errors and cannot obtain an ideal point cloud registration result. Summary of the invention
[0004] The purpose of the technical solution of the present invention is to provide a point cloud registration method, device, equipment, medium and program product, which are used to solve the problem that the ICP algorithm in the prior art easily falls into a local optimal solution when dealing with large posture errors and cannot obtain an ideal point cloud registration result.
[0005] To achieve the above object, the present invention is achieved by:
[0006] In a first aspect, an embodiment of the present invention provides a point cloud registration method, comprising:
[0007] Using genetic algorithm to obtain the rough registration parameters between the initial point cloud of the target workpiece and the model point cloud;
[0008] Generating a coarse registration point cloud of the target workpiece according to the initial point cloud and the coarse registration parameters;
[0009] According to the optimization target, obtaining the fine registration parameters between the coarse registration point cloud and the model point cloud;
[0010] The optimization objectives include minimizing the sum of the distances from each coarse registration point in the coarse registration point cloud to the corresponding near point in the model point cloud, and minimizing the distance between each coarse registration point and the corresponding near point.
[0011] Optionally, the point cloud registration method, wherein the rough registration parameters between the initial point cloud of the target workpiece and the model point cloud are obtained by using a genetic algorithm, comprises:
[0012] Initializing the coarse registration parameters between the initial point cloud and the model point cloud to random values to construct an initial population;
[0013] Performing natural selection, crossover operation and mutation operation on the initial population to generate a progeny population;
[0014] Obtaining the best individual with the smallest fitness function value in the offspring population;
[0015] According to the optimal individual, obtaining a coarse registration parameter between the initial point cloud and the model point cloud;
[0016] The fitness function value is the minimum average Euclidean distance of all point pairs between the initial point cloud and the model point cloud.
[0017] Optionally, in the point cloud registration method, obtaining fine registration parameters between the coarse registration point cloud and the model point cloud comprises:
[0018] For each coarse registration point in the coarse registration point cloud, obtaining a corresponding near point in the model point cloud, wherein the distance between each coarse registration point and the corresponding near point is the smallest;
[0019] The least square method or singular value decomposition method is used to obtain a fine registration parameter that minimizes the distance between each coarse registration point in the coarse registration point cloud and the corresponding near point in the model point cloud.
[0020] Optionally, the point cloud registration method further comprises:
[0021] Generating a fine registration point cloud of the target workpiece according to the coarse registration point cloud and the fine registration parameters;
[0022] Acquire a grinding allowance according to the model point cloud and the precise registration point cloud;
[0023] According to the grinding allowance, a trajectory of a grinding device when grinding the target workpiece is adjusted.
[0024] Optionally, the point cloud registration method further comprises:
[0025] Generating a fine registration point cloud of the target workpiece according to the coarse registration point cloud and the fine registration parameters;
[0026] Obtaining the registration accuracy of the precise registration point cloud;
[0027] If the registration accuracy is greater than or equal to the preset accuracy threshold, the step of using the genetic algorithm is repeated.
[0028] Optionally, the point cloud registration method, wherein obtaining the grinding allowance according to the model point cloud and the precise registration point cloud, comprises:
[0029] Obtaining the registration accuracy of the precise registration point cloud;
[0030] If the registration accuracy is less than a preset accuracy threshold, a grinding allowance is obtained according to the model point cloud and the precise registration point cloud.
[0031] In a second aspect, an embodiment of the present invention further provides a point cloud registration device, comprising:
[0032] A first acquisition module is used to acquire the rough registration parameters between the initial point cloud of the target workpiece and the model point cloud by using a genetic algorithm;
[0033] A first generating module, configured to generate a coarse registration point cloud of the target workpiece according to the initial point cloud and the coarse registration parameters;
[0034] A second acquisition module, used for acquiring a fine registration parameter between the coarse registration point cloud and the model point cloud according to an optimization target;
[0035] The optimization objectives include minimizing the sum of the distances from each coarse registration point in the coarse registration point cloud to the corresponding near point in the model point cloud, and minimizing the distance between each coarse registration point and the corresponding near point.
[0036] In a third aspect, an embodiment of the present invention further provides a point cloud registration device, comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the point cloud registration method as described in the first aspect is implemented.
[0037] In a fourth aspect, an embodiment of the present invention further provides a readable storage medium having a program stored thereon, and when the program is executed by a processor, the point cloud registration method as described in the first aspect is implemented.
[0038] In a fifth aspect, an embodiment of the present invention further provides a computer program product, comprising computer instructions, which, when executed by a processor, implement the point cloud registration method as described in the first aspect.
[0039] The beneficial effects of the above technical solution of the present invention are as follows:
[0040] The point cloud registration method described in the embodiment of the present invention is adopted, and the coarse registration parameters between the initial point cloud of the target workpiece and the model point cloud are obtained by using a genetic algorithm; the coarse registration point cloud of the target workpiece is generated according to the initial point cloud and the coarse registration parameters; the fine registration parameters between the coarse registration point cloud and the model point cloud are obtained according to the optimization target; in this way, the coarse registration parameters are obtained through the global search capability of the genetic algorithm during the point cloud registration process, and the ideal fine registration parameters are further obtained according to the initial alignment of the coarse registration parameters, thereby effectively avoiding local optimal solutions and improving the accuracy and robustness of point cloud registration. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1 A schematic diagram of a process of a point cloud registration method according to an embodiment of the present invention;
[0042] Figure 2 A schematic diagram of a flow chart of one implementation of the point cloud registration method according to an embodiment of the present invention;
[0043] Figure 3 A schematic diagram of a flow chart of another implementation of the point cloud registration method according to an embodiment of the present invention;
[0044] Figure 4 is a structural block diagram of the point cloud registration device according to an embodiment of the present invention;
[0045] Figure 5 Schematic diagram of the hardware structure of the point cloud registration device described in an embodiment of the present invention. DETAILED DESCRIPTION
[0046] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.
[0047] In various embodiments of the present invention, it should be understood that the size of the serial numbers of the following processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0048] Additionally, the terms "system" and "network" are often used interchangeably herein.
[0049] The terms "first", "second", etc. in the specification and claims of the present invention 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 where appropriate, 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" and "second" are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.
[0050] like Figure 1 As shown, an embodiment of the present invention provides a point cloud registration method, comprising:
[0051] S101, using a genetic algorithm to obtain coarse registration parameters between an initial point cloud of a target workpiece and a model point cloud.
[0052] In the embodiment of the present invention, the initial point cloud is the actual point cloud of the target workpiece; the model point cloud is the CAD design model point cloud of the target workpiece. The registration parameters are the registration results between the initial point cloud and the model point cloud of the target workpiece, and the coarse registration parameters are the initial registration results. The coarse registration parameters can also be understood as initial alignment parameters, including rotation parameters and translation parameters. The initial alignment parameters of the initial point cloud and the model point cloud are optimized using a genetic algorithm.
[0053] As a global search optimization algorithm, genetic algorithm can effectively avoid the problem of local optimal solution by simulating natural selection and genetic evolution process.
[0054] It should be noted that, before using the genetic algorithm to obtain the rough registration parameters between the initial point cloud of the target workpiece and the model point cloud in S101, the above method may further include:
[0055] Obtaining the initial point cloud and model point cloud of the target workpiece;
[0056] Preprocessing the initial point cloud and the model point cloud;
[0057] S101, using a genetic algorithm to obtain rough registration parameters between the initial point cloud of the target workpiece and the model point cloud, including:
[0058] Genetic algorithm is used to obtain the coarse registration parameters between the preprocessed initial point cloud and the model point cloud.
[0059] Furthermore, the initial point cloud and the model point cloud are preprocessed, and the preprocessing at least includes: noise reduction processing, downsampling processing and feature extraction processing.
[0060] Initial point cloud P = {p1, p2, ..., p m} and model point cloud Q = {q1,q2,…,q n}, where the workpiece point p i and model point q i It is a point in three-dimensional space. Noise points in the point cloud acquisition process can be removed through noise reduction processing to prevent noise interference. Downsampling processing can also be used to reduce the amount of data, avoid data redundancy, and improve subsequent processing efficiency.
[0061] S102: Generate a coarse registration point cloud of the target workpiece according to the initial point cloud and the coarse registration parameters.
[0062] S103, obtaining fine registration parameters between the coarse registration point cloud and the model point cloud according to the optimization target;
[0063] The optimization objectives include minimizing the sum of the distances from each coarse registration point in the coarse registration point cloud to the corresponding near point in the model point cloud, and minimizing the distance between each coarse registration point and the corresponding near point.
[0064] It can be understood that S102 and S103 are actually the ICP algorithm. In the embodiment of the present invention, the ICP algorithm is applied to achieve fine registration based on the coarse registration parameters obtained by the genetic algorithm. The ICP algorithm continuously iterates and calculates the closest point pairs between the initial point cloud and the model point cloud according to the optimization target, and minimizes the distance between these point pairs until it converges to the optimal solution.
[0065] The registration parameters are the registration results between the initial point cloud of the target workpiece and the model point cloud, and the fine registration parameters are the ideal registration results.
[0066] Wherein, S102, generating a coarse registration point cloud of the target workpiece according to the initial point cloud and the coarse registration parameters, is essentially a process of initializing the ICP algorithm.
[0067] It should be noted that there is a corresponding near point in the model point cloud that is closest to each coarse registration point in the coarse registration point cloud. The coarse registration point and its corresponding corresponding near point can be called a closest point pair.
[0068] The point cloud registration method described in the embodiment of the present invention can effectively improve the problem of low segmentation accuracy in the traditional Euclidean clustering algorithm, and is suitable for accurately segmenting workpiece point clouds in industrial scenes to identify workpieces.
[0069] In one implementation mode, optionally, using a genetic algorithm to obtain coarse registration parameters between an initial point cloud of a target workpiece and a model point cloud includes:
[0070] Initializing the coarse registration parameters between the initial point cloud and the model point cloud to random values to construct an initial population;
[0071] Performing natural selection, crossover operation and mutation operation on the initial population to generate a progeny population;
[0072] Obtaining the best individual with the smallest fitness function value in the offspring population;
[0073] According to the optimal individual, obtaining a coarse registration parameter between the initial point cloud and the model point cloud;
[0074] The fitness function value is the minimum average Euclidean distance of all point pairs between the initial point cloud and the model point cloud.
[0075] In an embodiment of the present invention, the point cloud registration problem is modeled as a parameter optimization problem, the coarse registration parameters are initialized to random values, and an initial population is constructed. Each individual can be represented by the coarse registration parameters (including rotation parameter R and translation parameter T).
[0076] Preferably, natural selection, crossover and mutation operations are performed on the initial population to generate a new generation of population, ie, offspring population.
[0077] Through multiple generations of evolutionary iterations, the optimal individual with the smallest fitness function value in the offspring population is found.
[0078] Here, the fitness function of the genetic algorithm is defined as the average Euclidean distance of all point pairs between the initial point cloud P and the model point cloud Q under the coarse registration parameters, and the formula is as follows:
[0079]
[0080] Among them, d ij =‖R·p i +Tq j ‖2, represents the workpiece point p i After rotation and translation, the model point q j The Euclidean distance between .
[0081] like Figure 2 The figure is a flowchart of one implementation of the point cloud registration method according to an embodiment of the present invention. According to this implementation, the method comprises the following steps:
[0082] S201, input the initial point cloud and the model point cloud.
[0083] S202, constructing an initial population.
[0084] S203, calculating the fitness function value of the initial population.
[0085] S204, determining whether the maximum number of iterations has been reached.
[0086] If the judgment result of S204 is yes (Y), then enter S205 to obtain the optimal individual according to the minimum fitness function value and save the optimal individual.
[0087] If the judgment result of S204 is no (N), the process proceeds to S206 to perform natural selection on the initial population.
[0088] S207, performing a crossover operation on the initial population after natural selection.
[0089] S208, performing mutation operation on the initial population after natural selection and crossover operation to generate a descendant population, and after S208, returning to S204.
[0090] In one implementation manner, optionally, obtaining fine registration parameters between the coarse registration point cloud and the model point cloud includes:
[0091] For each coarse registration point in the coarse registration point cloud, obtaining a corresponding near point in the model point cloud, wherein the distance between each coarse registration point and the corresponding near point is the smallest;
[0092] The least square method or singular value decomposition method is used to obtain a fine registration parameter that minimizes the distance between each coarse registration point in the coarse registration point cloud and the corresponding near point in the model point cloud.
[0093] In the embodiment of the present invention, for each coarse registration point p in the coarse registration point cloud, i ∈P, find the corresponding close point q in the model point cloud j ∈Q, such that:
[0094]
[0095] After obtaining the corresponding near point in the model point cloud for each coarse registration point in the coarse registration point cloud, the closest point pair is obtained.
[0096] It should be noted that, in the embodiment of the present invention, a kd tree data structure may be used for point cloud indexing, so that the nearest point pair can be obtained efficiently.
[0097] At the current iteration k, calculate the rotation parameter R (k) and the translation parameter T (k) Minimize the alignment error between the initial point cloud P and the model point cloud Q. The specific optimization goal is to minimize the following cost function:
[0098]
[0099] Use the least squares method or singular value decomposition method to solve the above formula and obtain the optimal rotation parameter R (k) and the translation parameter T (k) , that is, to obtain the precise registration parameters. Here, continuous iteration is required to ensure that the cost function E(R,T) converges to the preset tolerance range.
[0100] Furthermore, the fine registration parameter T (k) Update the initial point cloud of the target workpiece, that is, generate the precise registration point cloud of the target workpiece. Specifically, the following formula can be used:
[0101] p i (k+1) =R (k) ·p i (k) +T (k)
[0102] In one embodiment, optionally, the method further comprises:
[0103] Generating a fine registration point cloud of the target workpiece according to the coarse registration point cloud and the fine registration parameters;
[0104] Acquire a grinding allowance according to the model point cloud and the precise registration point cloud;
[0105] According to the grinding allowance, a trajectory of a grinding device when grinding the target workpiece is adjusted.
[0106] In the embodiment of the present invention, in order to register the initial point cloud of the target workpiece with its corresponding model point cloud, a fine registration point cloud of the target workpiece is generated according to the coarse registration point cloud and the fine registration parameters.
[0107] Calculate the difference between the precise registration point cloud of the target workpiece and the model point cloud to obtain the grinding allowance. The specific formula is:
[0108] Δ i =‖p i -q i ‖2
[0109] Among them, Δ i Represents workpiece point p i With model point q i The distance between the two points is the grinding allowance of the workpiece point.
[0110] For all workpiece points, calculate the average and maximum grinding allowances. The specific formula is:
[0111]
[0112] According to the above grinding allowance, the trajectory of the grinding equipment is adjusted to ensure that excess material is removed and the geometric accuracy is maintained.
[0113] In one embodiment, optionally, the method further comprises:
[0114] Generating a fine registration point cloud of the target workpiece according to the coarse registration point cloud and the fine registration parameters;
[0115] Obtaining the registration accuracy of the precise registration point cloud;
[0116] If the registration accuracy is greater than or equal to the preset accuracy threshold, the step of using the genetic algorithm is repeated.
[0117] In the embodiment of the present invention, in order to register the initial point cloud of the target workpiece with its corresponding model point cloud, a fine registration point cloud of the target workpiece is generated according to the coarse registration point cloud and the fine registration parameters.
[0118] According to the following formula, the registration accuracy of the precisely registered point cloud is obtained:
[0119]
[0120] Here, RMSE is the root mean square error. In the embodiment of the present invention, the root mean square error is used as the registration accuracy of the precise registration point cloud.
[0121] If the registration accuracy is greater than or equal to the preset accuracy threshold, it is considered that the current registration accuracy does not meet the polishing requirements and the genetic algorithm steps need to be repeated.
[0122] It should be noted that repeatedly utilizing the steps of the genetic algorithm means that steps S101 to S103 need to be executed again until a precise registration parameter is obtained that makes the registration accuracy less than a preset accuracy threshold.
[0123] In one implementation manner, optionally, obtaining a grinding allowance according to the model point cloud and the precise registration point cloud includes:
[0124] Obtaining the registration accuracy of the precise registration point cloud;
[0125] If the registration accuracy is less than a preset accuracy threshold, a grinding allowance is obtained according to the model point cloud and the precise registration point cloud.
[0126] The registration accuracy of the fine registration point cloud is obtained according to the formula in the above embodiment. If the registration accuracy is less than the preset accuracy threshold, it is considered that the current registration accuracy meets the polishing requirements, the fine registration parameters are saved, and the polishing task is performed according to the polishing allowance.
[0127] like Figure 3 This is a flow chart of another implementation of the point cloud registration method according to an embodiment of the present invention. In this implementation, the method includes the following steps:
[0128] S301, point cloud preprocessing: preprocess the initial point cloud and the model point cloud respectively, including noise reduction, downsampling and feature extraction processing.
[0129] S302, genetic algorithm optimization of initial alignment: construct an initial population through a genetic algorithm, use natural selection, crossover operation and mutation operation to evaluate the fitness function value of the individual, obtain the optimal individual, and then obtain the rough alignment parameters; the fitness function value is determined based on the Euclidean distance between the initial point cloud and the model point cloud.
[0130] S303, ICP algorithm fine registration: Based on the coarse registration parameters obtained by the genetic algorithm, the ICP algorithm is applied to continuously iterate the nearest point pairs between the initial point cloud and the model point cloud for fine registration until they converge to the optimal solution and obtain the fine registration parameters.
[0131] S304, allowance control: obtaining the grinding allowance according to the model point cloud and the precise registration point cloud.
[0132] S305, verification of registration result: verifying the registration accuracy of the finely registered point cloud by calculating the root mean square error between the finely registered point cloud and the model point cloud.
[0133] In summary, by adopting the point cloud registration method described in the embodiment of the present invention, combining the genetic algorithm with the ICP algorithm, the global search capability of the genetic algorithm can be used to optimize the initial alignment of the ICP algorithm during the registration process, thereby improving the registration accuracy and robustness to ensure the global optimality of the registration.
[0134] like Figure 4 As shown, an embodiment of the present invention further provides a point cloud registration device, comprising:
[0135] A first acquisition module 401 is used to acquire a rough registration parameter between an initial point cloud of a target workpiece and a model point cloud by using a genetic algorithm;
[0136] A first generating module 402, configured to generate a coarse registration point cloud of the target workpiece according to the initial point cloud and the coarse registration parameters;
[0137] A second acquisition module 403 is used to acquire a fine registration parameter between the coarse registration point cloud and the model point cloud according to an optimization target;
[0138] The optimization objectives include minimizing the sum of the distances from each coarse registration point in the coarse registration point cloud to the corresponding near point in the model point cloud, and minimizing the distance between each coarse registration point and the corresponding near point.
[0139] Optionally, in the point cloud registration device, the first acquisition module 401 is specifically used to:
[0140] Initializing the coarse registration parameters between the initial point cloud and the model point cloud to random values to construct an initial population;
[0141] Performing natural selection, crossover operation and mutation operation on the initial population to generate a progeny population;
[0142] Obtaining the best individual with the smallest fitness function value in the offspring population;
[0143] According to the optimal individual, obtaining a coarse registration parameter between the initial point cloud and the model point cloud;
[0144] The fitness function value is the minimum average Euclidean distance of all point pairs between the initial point cloud and the model point cloud.
[0145] Optionally, in the point cloud registration device, the second acquisition module 403 is specifically used to:
[0146] For each coarse registration point in the coarse registration point cloud, obtaining a corresponding near point in the model point cloud, wherein the distance between each coarse registration point and the corresponding near point is the smallest;
[0147] The least square method or singular value decomposition method is used to obtain a fine registration parameter that minimizes the distance between each coarse registration point in the coarse registration point cloud and the corresponding near point in the model point cloud.
[0148] Optionally, the point cloud registration device further comprises:
[0149] A second generating module, configured to generate a fine registration point cloud of the target workpiece according to the coarse registration point cloud and the fine registration parameters;
[0150] A third acquisition module is used to acquire a grinding allowance according to the model point cloud and the precise registration point cloud;
[0151] The adjustment module is used to adjust the trajectory of the grinding equipment when grinding the target workpiece according to the grinding allowance.
[0152] Optionally, the point cloud registration device further comprises:
[0153] A third generating module, configured to generate a fine registration point cloud of the target workpiece according to the coarse registration point cloud and the fine registration parameters;
[0154] A fourth acquisition module, used to obtain the registration accuracy of the precise registration point cloud;
[0155] A repetition module is used to repeat the step of using the genetic algorithm if the registration accuracy is greater than or equal to a preset accuracy threshold.
[0156] Optionally, in the point cloud registration device, the third acquisition module is specifically used to:
[0157] Obtaining the registration accuracy of the precise registration point cloud;
[0158] If the registration accuracy is less than a preset accuracy threshold, a grinding allowance is obtained according to the model point cloud and the precise registration point cloud.
[0159] The point cloud registration device provided in the embodiment of the present invention can execute the above-mentioned point cloud registration method embodiment, and its implementation principle and technical effect are similar, which will not be repeated in this embodiment.
[0160] like Figure 5As shown, an embodiment of the present invention also provides a point cloud registration device, including: a processor 501; and a memory 502 connected to the processor 501 through a bus interface, the memory 502 is used to store programs and data used by the processor 501 when performing operations, and the processor 501 calls and executes the programs and data stored in the memory 502.
[0161] The processor 501 is used to read the program in the memory 502 and execute the following process:
[0162] Using genetic algorithm to obtain the rough registration parameters between the initial point cloud of the target workpiece and the model point cloud;
[0163] Generating a coarse registration point cloud of the target workpiece according to the initial point cloud and the coarse registration parameters;
[0164] According to the optimization target, obtaining the fine registration parameters between the coarse registration point cloud and the model point cloud;
[0165] The optimization objectives include minimizing the sum of the distances from each coarse registration point in the coarse registration point cloud to the corresponding near point in the model point cloud, and minimizing the distance between each coarse registration point and the corresponding near point.
[0166] The transceiver 503 is used to receive and send data in the controller of the processor 501.
[0167] Among them, Figure 5 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 501 and various circuits of memory represented by memory 502 are linked together. The bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 503 may be a plurality of components, namely, a transmitter and a transceiver, providing a unit for communicating with various other devices on a transmission medium. For different user devices, the user interface 504 may also be an interface capable of externally and internally connecting required devices, and the connected devices include but are not limited to a keypad, a display, a speaker, a microphone, a joystick, and the like.
[0168] The processor 501 is responsible for managing the bus architecture and general processing, and the memory 502 can store data used by the processor 501 when performing operations.
[0169] Optionally, the processor 501 is specifically configured to read the computer program and execute the following steps:
[0170] Initializing the coarse registration parameters between the initial point cloud and the model point cloud to random values to construct an initial population;
[0171] Performing natural selection, crossover operation and mutation operation on the initial population to generate a progeny population;
[0172] Obtaining the best individual with the smallest fitness function value in the offspring population;
[0173] According to the optimal individual, obtaining a coarse registration parameter between the initial point cloud and the model point cloud;
[0174] The fitness function value is the minimum average Euclidean distance of all point pairs between the initial point cloud and the model point cloud.
[0175] Optionally, the processor 501 is specifically configured to read the computer program and execute the following steps:
[0176] For each coarse registration point in the coarse registration point cloud, obtaining a corresponding near point in the model point cloud, wherein the distance between each coarse registration point and the corresponding near point is the smallest;
[0177] The least square method or singular value decomposition method is used to obtain a fine registration parameter that minimizes the distance between each coarse registration point in the coarse registration point cloud and the corresponding near point in the model point cloud.
[0178] Optionally, the processor 501 is further configured to read the computer program and execute the following steps:
[0179] Generating a fine registration point cloud of the target workpiece according to the coarse registration point cloud and the fine registration parameters;
[0180] Acquire a grinding allowance according to the model point cloud and the precise registration point cloud;
[0181] According to the grinding allowance, a trajectory of a grinding device when grinding the target workpiece is adjusted.
[0182] Optionally, the processor 501 is further configured to read the computer program and execute the following steps:
[0183] Generating a fine registration point cloud of the target workpiece according to the coarse registration point cloud and the fine registration parameters;
[0184] Obtaining the registration accuracy of the precise registration point cloud;
[0185] If the registration accuracy is greater than or equal to the preset accuracy threshold, the step of using the genetic algorithm is repeated.
[0186] Optionally, the processor 501 is specifically configured to read the computer program and execute the following steps:
[0187] Obtaining the registration accuracy of the precise registration point cloud;
[0188] If the registration accuracy is less than a preset accuracy threshold, a grinding allowance is obtained according to the model point cloud and the precise registration point cloud.
[0189] A specific embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the above-mentioned point cloud registration method are implemented and the same technical effect can be achieved. To avoid repetition, it will not be described here.
[0190] In addition, an embodiment of the present invention further provides a computer program product, including computer instructions, which, when executed by a processor, implement the above Figure 1 The various processes of the method embodiment shown can achieve the same technical effect, and will not be described again here to avoid repetition.
[0191] In the several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0192] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may be physically included separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.
[0193] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform some steps of the sending and receiving methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store program codes.
[0194] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A point cloud registration method, characterized in that: include: Using genetic algorithm to obtain the rough registration parameters between the initial point cloud of the target workpiece and the model point cloud; Generating a coarse registration point cloud of the target workpiece according to the initial point cloud and the coarse registration parameters; According to the optimization target, obtaining the fine registration parameters between the coarse registration point cloud and the model point cloud; The optimization objectives include minimizing the sum of the distances from each coarse registration point in the coarse registration point cloud to the corresponding near point in the model point cloud, and minimizing the distance between each coarse registration point and the corresponding near point.
2. The point cloud registration method according to claim 1, characterized in that: Genetic algorithm is used to obtain the rough registration parameters between the initial point cloud of the target workpiece and the model point cloud, including: Initializing the coarse registration parameters between the initial point cloud and the model point cloud to random values to construct an initial population; Performing natural selection, crossover operation and mutation operation on the initial population to generate a progeny population; Obtaining the best individual with the smallest fitness function value in the offspring population; According to the optimal individual, obtaining a coarse registration parameter between the initial point cloud and the model point cloud; The fitness function value is the minimum average Euclidean distance of all point pairs between the initial point cloud and the model point cloud.
3. The point cloud registration method according to claim 1, characterized in that: Acquiring fine registration parameters between the coarse registration point cloud and the model point cloud, including: For each coarse registration point in the coarse registration point cloud, obtaining a corresponding near point in the model point cloud, wherein the distance between each coarse registration point and the corresponding near point is the smallest; The least square method or singular value decomposition method is used to obtain a fine registration parameter that minimizes the distance between each coarse registration point in the coarse registration point cloud and the corresponding near point in the model point cloud.
4. The point cloud registration method according to claim 1, characterized in that: Also includes: Generating a fine registration point cloud of the target workpiece according to the coarse registration point cloud and the fine registration parameters; Acquire a grinding allowance according to the model point cloud and the precise registration point cloud; According to the grinding allowance, a trajectory of a grinding device when grinding the target workpiece is adjusted.
5. The point cloud registration method according to claim 1, characterized in that: Also includes: Generating a fine registration point cloud of the target workpiece according to the coarse registration point cloud and the fine registration parameters; Obtaining the registration accuracy of the precise registration point cloud; If the registration accuracy is greater than or equal to the preset accuracy threshold, the step of using the genetic algorithm is repeated.
6. The point cloud registration method according to claim 4, characterized in that: Obtaining a grinding allowance according to the model point cloud and the precise registration point cloud, including: Obtaining the registration accuracy of the precise registration point cloud; If the registration accuracy is less than a preset accuracy threshold, a grinding allowance is obtained according to the model point cloud and the precise registration point cloud.
7. A point cloud registration device, characterized in that: include: A first acquisition module is used to acquire the rough registration parameters between the initial point cloud of the target workpiece and the model point cloud by using a genetic algorithm; A first generating module, used for generating a coarse registration point cloud of the target workpiece according to the initial point cloud and the coarse registration parameters; A second acquisition module, used for acquiring a fine registration parameter between the coarse registration point cloud and the model point cloud according to an optimization target; The optimization objectives include minimizing the sum of the distances from each coarse registration point in the coarse registration point cloud to the corresponding near point in the model point cloud, and minimizing the distance between each coarse registration point and the corresponding near point.
8. A point cloud registration device, characterized in that: include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the point cloud registration method according to any one of claims 1 to 6 is implemented.
9. A readable storage medium, characterized in that: The readable storage medium stores a program, and when the program is executed by a processor, the point cloud registration method according to any one of claims 1 to 6 is implemented.
10. A computer program product, characterized in that The method comprises computer instructions, which, when executed by a processor, implement the point cloud registration method according to any one of claims 1 to 6.