Unmanned aerial vehicle point cloud data-based heap volume measurement method, apparatus and device, and storage medium

Abstract

The invention discloses a pile volume measurement method, device and equipment based on unmanned aerial vehicle point cloud data and a storage medium, and relates to the technical field of point cloud data processing. According to the scheme, at a Web end, original point cloud data collected by an unmanned aerial vehicle are classified through a preset deep learning model constructed based on a PointNet + + network, and a target heap volume corresponding to target point cloud data is calculated through an Alpha Shapes algorithm, so that the calculation efficiency and the calculation precision are improved; and meanwhile, the unmanned aerial vehicle executes a point cloud scanning acquisition task, so that a measurement position can be effectively reached, and the problem that a target object cannot be completely covered can be solved.

Description

Technical Field

[0001] This application relates to the field of point cloud data processing technology, and in particular to a method, apparatus, equipment and storage medium for measuring the volume of a stack based on UAV point cloud data. Background Technology

[0002] The measurement of the volume of a stockpile has important research significance and application value in various engineering projects, such as geological surveying, building surveying, port stacking, and mining exploration.

[0003] In related technologies, traditional volume measurement methods (such as regular grid methods and voxelization methods) typically rely on dense sampling or rasterization processing, resulting in low computational efficiency and insufficient accuracy. Furthermore, traditional volume measurement methods, such as backpack-mounted or vehicle-mounted LiDAR scanners, suffer from insufficient measurement height, blind spots, and inability to completely cover targets of a certain height (such as port cargo stacks and urban buildings). Moreover, for application scenarios such as underground mines, cross-river bridges, and overhead power transmission lines, backpack-mounted or vehicle-mounted LiDAR scanners cannot reach the effective measurement location. Summary of the Invention

[0004] This specification provides an embodiment of a method for measuring the volume of a stack based on UAV point cloud data, in order to solve the problems of low computational efficiency, insufficient accuracy, and inability to effectively scan target objects in traditional volume measurement methods in the prior art.

[0005] To solve the above-mentioned technical problems, the embodiments in this specification are implemented as follows:

[0006] Firstly, the method for measuring the volume of a stack based on UAV point cloud data provided in the embodiments of this specification, when applied to a web-based application, includes:

[0007] Acquire raw point cloud data of the preset inspection area collected by the drone;

[0008] The original point cloud data is classified and target point cloud data is identified based on a preset deep learning model; the preset deep learning model is built based on the PointNet++ network.

[0009] The target heap volume corresponding to the target point cloud data is calculated based on the Alpha Shapes algorithm.

[0010] Secondly, the heap volume measurement device based on UAV point cloud data provided in the embodiments of this specification, applied to a web platform, includes:

[0011] The acquisition module is used to acquire raw point cloud data of the preset inspection area collected by the UAV;

[0012] The identification module is used to classify the original point cloud data based on a preset deep learning model and identify target point cloud data; the preset deep learning model is built based on the PointNet++ network.

[0013] The determination module is used to calculate the target heap volume corresponding to the target point cloud data based on the Alpha Shapes algorithm.

[0014] Thirdly, the stack volume measurement device based on UAV point cloud data provided in the embodiments of this specification includes a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the stack volume measurement method based on UAV point cloud data in Scheme 1.

[0015] Fourthly, the embodiments of this specification provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the stack volume measurement method based on UAV point cloud data in Scheme 1.

[0016] One embodiment of this specification achieves the following beneficial effects: On the Web, the raw point cloud data collected by the UAV is classified by a preset deep learning model built on the PointNet++ network, and the target pile volume corresponding to the target point cloud data is calculated by the AlphaShapes algorithm, which improves the calculation efficiency and accuracy. At the same time, the UAV can effectively reach the measurement position and solve the problem of not being able to completely cover the target object when performing point cloud scanning and acquisition tasks. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the method for measuring the volume of a stack based on UAV point cloud data provided in the embodiments of this specification;

[0019] Figure 2 A schematic diagram of the raw point cloud data provided in the embodiments of this specification;

[0020] Figure 3 A schematic diagram of the classification point cloud data provided in the embodiments of this specification;

[0021] Figure 4 A schematic diagram of effective point cloud data provided in the embodiments of this specification;

[0022] Figure 5 A schematic diagram of the structure of the stack volume measurement device based on UAV point cloud data provided in the embodiments of this specification;

[0023] Figure 6 This is a schematic diagram of the structure of a stack volume measurement device based on UAV point cloud data provided in the embodiments of this specification. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of one or more embodiments of this specification clearer, the technical solutions of one or more embodiments of this specification will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of them. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of one or more embodiments of this specification.

[0025] The technical solutions provided in the various embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0026] The method for measuring the volume of a stack based on UAV point cloud data, as provided in the embodiments of the specification, will be described in detail with reference to the accompanying drawings.

[0027] Figure 1 This is a flowchart illustrating the method for measuring the volume of a stack based on UAV point cloud data, as provided in the embodiments of this specification.

[0028] like Figure 1 As shown, this application, used on a web platform, can process point cloud data in real time and efficiently. The process may include the following steps:

[0029] Step 110: Obtain the raw point cloud data of the preset inspection area collected by the drone.

[0030] In this embodiment, the UAV is equipped with a LiDAR, takes off from an automated airport, scans a preset inspection area along a preset route, generates a point cloud in LAS1.4 format in real time, and transmits it back to the edge server via a 5G private network. The UAV can fly at an altitude of 50m and a speed of 8m / s. The LiDAR parameters can be set to a scanning frequency of 240Hz, a point density of 150pt / ㎡, and RTK positioning of 1cm+1ppm horizontally and 2cm+1ppm vertically. The inspection area can be a bulk cargo terminal, urban buildings, etc.

[0031] Figure 2 This is a schematic diagram of the raw point cloud data provided in the embodiments of this specification.

[0032] Taking bulk cargo terminal yards as an example, such as Figure 2 As shown, a point cloud in LAS1.4 format is generated by scanning with a lidar mounted on a drone.

[0033] Step 120: Classify the original point cloud data based on a preset deep learning model and identify the target point cloud data; the preset deep learning model is built based on the PointNet++ network.

[0034] In the embodiments described in this specification, the preset deep learning model is built based on an optimized PointNet++ network, and the preset deep learning model is used to achieve automatic classification of point cloud data.

[0035] Step 130: Calculate the target heap volume corresponding to the target point cloud data based on the Alpha Shapes algorithm.

[0036] In the embodiments of this specification, the calculation of the target heap volume corresponding to the target point cloud data based on the Alpha Shapes algorithm is a process that combines surface reconstruction and spatial segmentation.

[0037] The automated drone airport enabled efficient collection and transmission of point cloud data. Combined with deep learning and the AlphaShapes algorithm, it enabled automated, high-precision pile volume measurement on the web.

[0038] It should be understood that the order of some steps in the methods described in one or more embodiments of this specification may be interchanged according to actual needs, or some steps may be omitted or deleted.

[0039] In the embodiments of this specification, on the Web side, the raw point cloud data collected by the UAV is classified by a preset deep learning model built on the PointNet++ network. The Alpha Shapes algorithm is used to calculate the target pile volume corresponding to the target point cloud data, which improves the calculation efficiency and accuracy. At the same time, the UAV can effectively reach the measurement position and solve the problem of not being able to completely cover the target object when performing point cloud scanning and acquisition tasks.

[0040] based on Figure 1 In addition to the method described in the embodiments of this specification, some specific implementation schemes of the method are also provided, which will be described below.

[0041] Optionally, the classification of the original point cloud data and identification of target point cloud data based on a preset deep learning model described in the embodiments of this specification may specifically include:

[0042] The farthest point sampling algorithm is used to select key points from the original point cloud data and construct a local neighborhood for each key point;

[0043] Calculate the relative position encoding of the neighboring points within the local neighborhood, and extract local features through a self-attention mechanism;

[0044] By using multi-layer downsampling and feature aggregation, the local features are integrated into the global features;

[0045] The global features are classified using a fully connected neural network to identify target point cloud data.

[0046] In the embodiments described in this specification, key points are selected from the original point cloud data to ensure that the key points are evenly distributed and cover the entire point cloud space. For example, M key points are selected from the input original point cloud data P, forming a key point set. Among them, S j For the j-th key point, the S-th key point j The key point sampling set of layer +1 is

[0047] For each key point S j Local neighborhood N(S) is constructed by grouping through ball queries. j ), Where r is the neighborhood radius, which can be adjusted according to actual needs (e.g., r = 0.1m); p k Let be the three-dimensional coordinate vector of the k-th point within the local area.

[0048] For each neighborhood point p in the local neighborhood k Calculate its relative to the key point S j coordinate offset Δp k , △p k =p k -s j e k =MLP(△p k The system generates relative positional codes using a multilayer perceptron (MLP). Local features within a local neighborhood are extracted through query, key, and value generation and attention weight calculation.

[0049] Q k =W q [f k ||e k ], K k =W k [f k ||e k ], V k =W v [f k ||e k ], where f k e represents the original feature of the k-th point. k To encode the position of the k-th point, Wq W K W V All are learnable weights, and || indicates concatenation.

[0050] In practice, for a neighboring point p k and p l Calculate its attention weight A kl Attention weights reflect the spatial correlation between them.

[0051] Aggregate neighborhood information based on attention weights to generate a new feature f′ for each point. k ,

[0052] Repeat the farthest point sampling and local feature extraction steps, gradually reducing the number of point clouds (downsampling up to layer L). Each layer outputs a set of key points and corresponding features, which are then aggregated through multi-layer downsampling and feature aggregation to form hierarchical global features. In each layer, neighborhood features are aggregated using max pooling or self-attention mechanisms to generate higher-level feature representations. For keypoint S... j The neighborhood features are aggregated to output a D-dimensional vector.

[0053] Max pooling is performed on the global features of the final layer to generate a fixed-dimensional global feature vector. A fully connected neural network (MLP) is then used to map the global features to a class space, outputting the probability distribution of each point belonging to each class, thus identifying the target point cloud data. (Class probability vector) Among them, f global A global maximization pooling method is used, and Softmax is a normalized exponential function. Cross-entropy loss is employed, and model parameters are optimized through backpropagation to improve classification accuracy. Loss function Among them, y c One-hot encoding of the true category (y c =1 indicates the true class, the rest are 0), C is the total number of classes (e.g., C=40 in ModelNet40), Let be the class probability vector of class c.

[0054] Figure 3 This is a schematic diagram of the classification point cloud data provided in the embodiments of this specification.

[0055] like Figure 3 As shown, the raw point cloud data can be categorized into point clouds of cargo stacks, point clouds of construction vehicles, point clouds of high-mast lights, point clouds of buildings, point clouds of dust, etc., with different colors representing different types of point cloud data. Identifying the point clouds of cargo stacks allows for the calculation of the volume of the stack.

[0056] Optionally, the calculation of the target heap volume corresponding to the target point cloud data based on the Alpha Shapes algorithm described in the embodiments of this specification may specifically include:

[0057] Calculate the point density of each single point cloud data in the target point cloud data within a preset radius sphere, and generate adaptive parameters for the target point cloud data;

[0058] Alpha Shapes surface reconstruction is performed based on the adaptive parameters;

[0059] The Delaunay triangulation method is applied to the boundary polygons extracted by Alpha Shapes to generate triangular meshes;

[0060] Based on the triangular mesh, the target pile volume corresponding to the target point cloud data is calculated.

[0061] In this embodiment of the specification, for each point pi in the target point cloud data, a spherical region is delineated with a preset radius r (e.g., r = 1m), the number of points within this spherical region is counted, and the point density is calculated. A mapping relationship between the point density and the Alpha parameter α is established. Wherein, α(p i ) represents the Alpha parameter of the region where point pi is located, α base =0.3m (reference value, calibrated experimentally), ρ local (pi) represents the point density of point pi within a sphere of radius 1m, ρ glocal This represents the point density within the local neighborhood.

[0062] For each point pi, an Alpha Shape is constructed using its corresponding adaptive parameter αi. The neighborhood range of point pi is determined based on αi, and points within the neighborhood that may constitute the surface are selected. Local surface fragments are constructed by connecting neighboring points to form triangles. All local surface fragments are merged to generate a continuous 3D surface model.

[0063] Closed boundary polygons are extracted from the Alpha Shapes surface model. Delaunay triangulation CDT is performed on the boundary polygons to ensure that the triangulation meets geometric constraints. The boundary polygons are projected onto a two-dimensional plane. The CDT algorithm is called on the two-dimensional plane to generate an optimized triangulation structure. The two-dimensional triangulation structure is stretched into three-dimensional space to restore its original height information and form a closed triangular mesh.

[0064] Combining the triangular mesh and the reference plane elevation (obtained through RANSAC plane fitting), layered volume calculations are performed. The reference plane elevation is determined by fitting the ground plane using the RANSAC algorithm. The triangular mesh is then divided into multiple thin layers according to height, each layer being treated as a triangular prism or frustum. The volume of each layer is calculated, and the volumes of all layers are summed to obtain the total volume of the target pile. Total Volume Among them, A k Let h be the projected area of ​​the k-th surface triangle. base The reference elevation (obtained through RANSAC plane fitting) The internal voxel unit volume (5cm resolution) is represented by h1, h2, and h3, which are the heights of the three vertices of the triangle, respectively. K is the number of a single triangle, and M is the total number of triangles.

[0065] In practice, by adapting to different regional densities through dynamic α parameters, the boundary error is reduced by 61%. Combined with KD trees and voxelization, it is 22% faster than traditional methods. It supports LAS / PLY format input and outputs standardized volume reports.

[0066] Optionally, before classifying the original point cloud data based on a preset deep learning model as described in the embodiments of this specification, the method may include:

[0067] The raw point cloud data is preprocessed, including denoising and / or registration.

[0068] The preprocessed raw point cloud data is filtered based on spatial indexing to obtain valid point cloud data.

[0069] In the embodiments of this specification, the original point cloud data may contain noise points due to LiDAR errors, environmental interference (or object reflection characteristics), which may manifest as isolated points, outliers, or non-target object points. To reduce the interference of noise points on the preset deep learning model, denoising processing can be performed. The point cloud space is divided into a 10m×10m×10m cubic voxel grid. Each voxel is traversed, the number of points inside is counted, and then the density is calculated. When the density is less than 1 pt, the current outlier, error point, isolated point, or floating point is marked as a noise point.

[0070] When raw point cloud data is acquired from different perspectives, registration is required to unify the multi-view point clouds into the same coordinate system and eliminate errors in overlapping areas. Registration is performed using ground features (fitted to the ground using RANSAC).

[0071] The point cloud filtering method optimized using spatial indexing (R-tree) obtains effective point cloud data based on the input polygonal region and height range (minHeight-maxHeight), reducing the input size of subsequent preset deep learning models.

[0072] Valid point cloud dataset P valid ={p i ∈P|p i ∈Ω,h min ≤z i ≤h max}, where P is the original point cloud set, containing all scanned point cloud data P i =(x i y i , z i ); Ω represents the polygonal region interactively defined by the user (such as the boundary of a port cargo stack); h min h max For height range constraints, used to filter out invalid points (such as ground or high-altitude noise).

[0073] Spatial index (R-tree) optimizes and speeds up queries: P candidate =R-treeQuery(P,MBR(Ω)), where MBR(Ω) is the minimum bounding rectangle of region Ω; R-treeQuery is a spatial index query function that returns the set of candidate points falling within the MBR. For example, filtering a target region (50m×50m) from a 2 million point cloud takes only 0.3s with an R-tree query.

[0074] Figure 4 This is a schematic diagram of the effective point cloud data provided in the embodiments of this specification.

[0075] like Figure 4 As shown, the effective point cloud data is divided into two types: the red part represents the point cloud data of the ground, and the green part represents the point cloud data of the stacks of goods or other items on the ground.

[0076] By reasonably reducing data density, the amount of computation can be significantly reduced. Furthermore, optionally, the method described in the embodiments of this specification may include:

[0077] The effective point cloud data is thinned according to the thinning factor.

[0078] In the embodiments described in this specification, the three-dimensional space is divided into a regular voxel grid according to a thinning factor. For each non-empty voxel, the centroid of all points within it is calculated (or a point is randomly selected) as the representative point of that voxel. All representative points are merged to form the thinned point cloud data. The thinned point cloud dataset P decimated ={p i ∈P valid |i≡0(mod d)}, where d is the thinning factor (e.g., d=10 means keeping 1 point out of every 10 points), i is the index of the point in the point cloud sequence, and P valid This refers to valid point cloud data within the target area.

[0079] In practice, in the measurement of coal stockpiles at port bulk cargo terminals, the processing time was reduced from 46 minutes to 18 minutes using the traditional volume measurement method, with an error of less than 1% compared to manual measurement results. Five edge depression areas that were missed by the traditional method were successfully identified.

[0080] Figure 5 This is a schematic diagram of the structure of the stack volume measurement device based on UAV point cloud data provided in the embodiments of this specification.

[0081] Corresponding to the method embodiment, this embodiment also provides a heap volume measurement device based on UAV point cloud data, which may include:

[0082] The acquisition module 502 is used to acquire the raw point cloud data of the preset inspection area collected by the UAV;

[0083] The identification module 504 is used to classify the original point cloud data based on a preset deep learning model and identify target point cloud data; the preset deep learning model is built based on the PointNet++ network.

[0084] The determination module 506 is used to calculate the target heap volume corresponding to the target point cloud data based on the Alpha Shapes algorithm.

[0085] Optionally, the classification of the original point cloud data and identification of target point cloud data based on a preset deep learning model described in the embodiments of this specification may specifically include:

[0086] The farthest point sampling algorithm is used to select key points from the original point cloud data and construct a local neighborhood for each key point;

[0087] Calculate the relative position encoding of the neighboring points within the local neighborhood, and extract local features through a self-attention mechanism;

[0088] By using multi-layer downsampling and feature aggregation, the local features are integrated into the global features;

[0089] The global features are classified using a fully connected neural network to identify target point cloud data.

[0090] Optionally, the calculation of the target heap volume corresponding to the target point cloud data based on the Alpha Shapes algorithm described in the embodiments of this specification may specifically include:

[0091] Calculate the point density of each single point cloud data in the target point cloud data within a preset radius sphere, and generate adaptive parameters for the target point cloud data;

[0092] Alpha Shapes surface reconstruction is performed based on the adaptive parameters;

[0093] The Delaunay triangulation method is applied to the boundary polygons extracted by Alpha Shapes to generate triangular meshes;

[0094] Based on the triangular mesh, the target pile volume corresponding to the target point cloud data is calculated.

[0095] Based on the same idea, this specification also provides devices corresponding to the above methods in its embodiments.

[0096] Figure 6 This is a schematic diagram of the structure of a stack volume measurement device based on UAV point cloud data provided in an embodiment of this specification. Figure 6 As shown in the embodiments of this specification, the stack volume measurement device 600 based on UAV point cloud data includes a memory 630, a processor 610, and a computer program 620 stored in the memory. The processor 610 executes the computer program 620 to implement the stack volume measurement method based on UAV point cloud data described in any of the above embodiments.

[0097] The stack volume measurement device based on UAV point cloud data provided in the embodiments of this specification may include a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the stack volume measurement method based on UAV point cloud data described in any of the above embodiments.

[0098] This specification provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the stack volume measurement method based on UAV point cloud data described in any of the above embodiments.

[0099] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, for... Figure 6 As the device shown is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.

[0100] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many methodological improvements today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that a methodological improvement cannot be implemented using hardware physical modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program and "integrate" a digital system onto a PLD themselves, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should understand that by simply performing some logic programming on the method flow using one of these hardware description languages ​​and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.

[0101] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, ASICs, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0102] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0103] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0104] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0105] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0106] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0107] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0108] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0109] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0110] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0111] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0112] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0113] This application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0114] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for measuring the volume of a stack based on UAV point cloud data, characterized in that, Applied to the web, including: Acquire raw point cloud data of the preset inspection area collected by the drone; The original point cloud data is classified and target point cloud data is identified based on a preset deep learning model; the preset deep learning model is built based on the PointNet++ network. The target heap volume corresponding to the target point cloud data is calculated based on the Alpha Shapes algorithm.

2. The method according to claim 1, characterized in that, The process of classifying the raw point cloud data and identifying target point cloud data based on a preset deep learning model specifically includes: The farthest point sampling algorithm is used to select key points from the original point cloud data and construct a local neighborhood for each key point; Calculate the relative position encoding of the neighboring points within the local neighborhood, and extract local features through a self-attention mechanism; By using multi-layer downsampling and feature aggregation, the local features are integrated into the global features; The global features are classified using a fully connected neural network to identify target point cloud data.

3. The method according to claim 1, characterized in that, The calculation of the target heap volume corresponding to the target point cloud data based on the Alpha Shapes algorithm specifically includes: Calculate the point density of each single point cloud data in the target point cloud data within a preset radius sphere, and generate adaptive parameters for the target point cloud data; Alpha Shapes surface reconstruction is performed based on the adaptive parameters; The Delaunay triangulation method is applied to the boundary polygons extracted by Alpha Shapes to generate triangular meshes; Based on the triangular mesh, the target pile volume corresponding to the target point cloud data is calculated.

4. The method according to claim 1, characterized in that, Before classifying the original point cloud data based on a preset deep learning model, the method includes: The raw point cloud data is preprocessed, including denoising and / or registration. The preprocessed raw point cloud data is filtered based on spatial indexing to obtain valid point cloud data.

5. The method according to claim 4, wherein the method comprises: The effective point cloud data is thinned according to the thinning factor.

6. A device for measuring the volume of a stack based on UAV point cloud data, characterized in that, Applied to the web, including: The acquisition module is used to acquire raw point cloud data of the preset inspection area collected by the UAV; The identification module is used to classify the original point cloud data based on a preset deep learning model and identify target point cloud data; the preset deep learning model is built based on the PointNet++ network. The determination module is used to calculate the target heap volume corresponding to the target point cloud data based on the Alpha Shapes algorithm.

7. The apparatus according to claim 6, characterized in that, The process of classifying the raw point cloud data and identifying target point cloud data based on a preset deep learning model specifically includes: The farthest point sampling algorithm is used to select key points from the original point cloud data and construct a local neighborhood for each key point; Calculate the relative position encoding of the neighboring points within the local neighborhood, and extract local features through a self-attention mechanism; By using multi-layer downsampling and feature aggregation, the local features are integrated into the global features; The global features are classified using a fully connected neural network to identify target point cloud data.

8. The apparatus according to claim 6, characterized in that, The calculation of the target heap volume corresponding to the target point cloud data based on the Alpha Shapes algorithm specifically includes: Calculate the point density of each single point cloud data in the target point cloud data within a preset radius sphere, and generate adaptive parameters for the target point cloud data; Alpha Shapes surface reconstruction is performed based on the adaptive parameters; The Delaunay triangulation method is applied to the boundary polygons extracted by Alpha Shapes to generate triangular meshes; Based on the triangular mesh, the target pile volume corresponding to the target point cloud data is calculated.

9. A device for measuring the volume of a stack based on UAV point cloud data, comprising a memory, a processor, and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 5.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 5.

Images