Research On Rule Construction Method Of Single Wood L-system Based On 3D Point Cloud Skeleton Extraction

The current wave of virtual reality technology has boosted the development and prosperity of 3D modeling technology.As a representative of natural landscape models,trees play an imperative role in virtual modeling research.Due to the wide variety of trees and the large differences in geometry between classes,they cannot be modeled in batches.At the same time,with the improvement of the needs,new requirements for the timeliness and authenticity of modeling are also put forward.The L system is a simple and highly structured fractal modeling method.This method with many variants can be modeled according to the constructed grammar rules,which are applicable to various modeling requirements.However,traditional L systems need to use relevant mathematical foundations and expertise to build rules,which are not conducive to fast and accurate modeling of large scenes due to high cost.In order to solve the above problems,this paper proposes an adaptive rule extraction method for L system.The method uses point cloud data to extract the parameter information of the tree skeleton and branches,and then uses the obtained skeleton to extract the non-parametric production of trees.Finally,the parameters are matched with the production formula to obtain the L system rules corresponding to the trees,and the resulting rules are used for modeling.The accuracy of the resulting rule is verified by comparing the similarity between the model and the real object.This article has mainly done several things:1)Firstly,the current ideas and methods based on L system modeling are studied.By systematically combing the literature,it summarizes the progress and shortcomings of the L system in the field of plant modeling in recent years.At the same time,the factors affecting the construction of the rules of the L system were analyzed.2)In order to efficiently operate the basic massive point cloud data,a kind of octree mixed point cloud index structure is proposed.This section briefly describes the principles of KD and octree;the octree-like ideas and coding methods are described in detail and verified by experiments.3)An optimization algorithm is proposed for the L1 skeleton extraction algorithm to solve the problem of insufficient skeleton accuracy due to missing point clouds and poor skeleton repeatability caused by random sampling.Firstly,based on the octree-based spatial segmentation,the adaptive point cloud enhancement is performed on the model with severe local missing;then the sub-space-based nearest neighbor sampling algorithm is used to downsample the enhanced point cloud;Finally,the sampling points are used to extract the skeleton in the enhanced point cloud.4)In order to obtain a complete L system rule,firstly uses the analytic algorithm to parse the skeleton,and obtains the topology and branch parameters for storage with a specific data structure;Next,the complete L system rule is obtained by matching the no-argument production with the parameters using a rule generation algorithm.In order to verify the validity of the resulting rules,we import the generated L-system rules into the L-studio software,and verify the validity of the generation rules by generating a simulation model and rea l tree comparison.