Human Body Static Modeling Based On Artificial Bee Colony Point Cloud Registration

In recent years,due to the continuous development and progress of computer vision technology,three-dimensional human body modeling has continued to flourish in the fields of medical imaging,industrial modeling,and intelligent robots.In the modeling process of the three-dimensional human body model,the three-dimensional registration technology is the most critical existence.The accuracy of registration is directly affected by the quality of point cloud data and the accuracy of feature points.When scanning to obtain point cloud data,noise points in the point cloud data are bound to exist due to the influence of the scanning environment and the device itself.In the registration process,due to the diversity of feature points and the problem of the matching rate,more accurate feature point pairs are often not obtained,which directly affects the post-registration effect and reduces the model's realism.Therefore,it is necessary to denoise the point cloud data and accurately extract the feature point pairs,and finally obtain a more accurate three-dimensional human body model.Aiming at the problem of high point cloud accuracy requirements in 3D modeling,this paper proposes a denoising step that combines the filtering denoising algorithm and the two-step denoising method of the least squares algorithm.The use of filtering denoising can remove the model edges well.Detect and scan some noise points in the background of the model,filter and denoise for the first time,and then combine the least square method to perform secondary denoising.Least squares can further filter the noise points,remove some noise points that are useless for registration and ensure the basic features of the human body model to a greater extent,which greatly improves the efficiency of point cloud registration in the later stage and shortens the registration.Accurate time.In view of the long time to find feature points in the registration process,this paper first uses the curvature information of the midpoint of the point cloud micro-cut plane to distinguish the concave and convex points,and compares the concave and convex points with the threshold to find the feature points in the model.It greatly reduces the number of feature points that need to be registered in the later stage,shortens the registration time and calculation cost,and provides better feature points for the later artificial bee colony algorithm registration.Aiming at the problem of requiring high-precision feature points in the registration process,this paper proposes a human body modeling technology based on artificial bee colony algorithm registration,which improves the algorithm through artificial bee colony and uses the cosine of the normal vector of the feature point cloud.The value tending to 1 is used as the objective function in the bee colony algorithm,so as to continuously iteratively obtain the optimal matching feature points in the point cloud to be registered,and obtain the optimal transformation matrix to drive the registration of the overall model.This algorithm is extremely It greatly reduces the influence of incorrect feature points on the model registration,thus completing the establishment of the model.In this paper,the human point cloud model is used to carry out registration experiments.The final experimental results show that the two-step denoising method proposed in this paper can well reduce the influence of outlier noise reduction on late registration.Curvature information is used to find the feature points and combine with the objective function of artificial bees to find the optimal pair of feature points.Finally,the registration and establishment of human body model are realized.By comparing with other registration modeling algorithms,the method proposed in this paper can achieve better registration effect and finally complete human body registration modeling.