3D Point Cloud Reconstruction Based On Fusion Improved Feature Matching And Enhanced ICP

In the processing of fine small objects,usually a three-dimensional point cloud model of the object to be processed is established at first,and then the more refined industrial processing work is completed on the basis of this model.However,when constructing a three-dimensional point cloud model,problems such as lack of local structural details and insufficient overall reconstruction accuracy will occur.In order to solve such problems and improve the accuracy of 3D point cloud reconstruction,this paper designs and proposes a 3D point cloud reconstruction system based on fusion,improved feature matching and enhanced ICP algorithm.The main works of the paper include:1.In the feature extraction matching link,this paper proposes an enhanced feature extraction matching algorithm.First,a multi-scale adaptive FAST algorithm and SURF algorithm proposed in this paper are used for fusion feature extraction.Then,the characteristic point direction and the description sub-calculation are determined by haar wavelet response.Finally,the acceleration KNN two-way matching and the RANSAC algorithm of this paper are used to complete the matching.Compared with other feature extraction and matching algorithms,the overall effective matching number of the algorithm in this paper is more,the matching efficiency reaches 93%,and it has good real-time and robustness.At the same time,it provides a more accurate initial value of the rotation and translation matrix for the 3D reconstruction system in this paper,and improves the reconstruction effect of the 3D point cloud model.2.In the point cloud fusion reconstruction link,this paper proposes an enhanced ICP algorithm.First,calculate the transformation parameters of each frame aligned to the world coordinate system through a large number of feature points between the matched different frames as the accurate initial value.Then adopt different weight selection strategies for the successfully matched and unmatched 3D point sets.Finally,a two-layer iterative optimization strategy proposed in this paper is used for point cloud fusion reconstruction.Experiments have proved that compared with traditional ICP,KC,RPM,and CPD algorithms,the algorithm in this paper has higher accuracy,with an average reconstruction error of about 0.077 mm,simple operation,low modeling cost,and good real-time performance.It can be applied to daily small The object’s 3D point cloud model reconstruction and experimental teaching.It can be used in 3D point cloud model reconstruction of small objects and experimental teaching.