A Novel Thinning Algorithm of 3D Image Model Based on Spatial Wavelet Interpolation

In dynamic modeling process of 3D image, too much few characteristics extracted in the binocular stereo vision modeling process cause the cloud characteristics of the model node could not be recovered and low fidelity of the model. In order to solve this problem, this paper presents a thinning algorithm of 3D image model based on improved spatial wavelet interpolation. Firstly, several coordinates of dynamic characteristic nodes are collected by the binocular stereo vision method to compose a characteristic' assemblage which is used to build an interpolation region. Then the characteristic nodes of interpolation are built in the region by the dynamic wavelet recovering method, and modified in direction. Whether the characteristic is reasonable is judged by the modified result. Finally the new characteristic is taken as the vertex to construct a new region and complete the iterative interpolation. Simulation results show that the accuracy error of the interpolation is less than 15 pixels when image interpolation is carried out based on a few characteristics which basically meet the requirements of stability, accuracy and anti-interference ability of 3D interpolation.