Reconstruction and Processing of Scene Point Cloud Based on Video

Abstract

This paper will construct augmented reality assembly system as the goal, the 3D reconstruction efficiency is low, the late recognition and perception problems, propose a scenario based reconstruction algorithm of video sequence, 3D sparse surface structure to recover the scene from image sequences, improve operational efficiency, ensure the accuracy and robustness of the reconstruction results. According to the characteristics of visual reconstruction noise, filtering method for filtering point clouds reconstruction using statistical outliers, filtering a large amount of noise without increasing the computational cost of many conditions, reduce the number of space and enhance the surface characteristics.