Boundary Incremental Reconstruction of the Known Space and Unknown Space in Indoor Scenes Based on Surfels

For the three-dimensional reconstruction of indoor scenes, to solve the problems of low point cloud reconstruction accuracy and high space complexity, a surfel-based incremental reconstruction algorithm for the boundary between the known space and the unknown space of the indoor scene is proposed. The algorithm uses the Kinect v1 depth camera to collect actual scene datasets and combines them with public datasets to generate a set of observed surfels, and simulates the boundary between a known space and an unknown space. The algorithm calculates the Boolean union of two bounded volumes, fusing the known volume of the previous frame with the space observed by the current depth image to generate a dense point cloud of the object. The algorithm proposed is effective and advanced, has important promotion significance for the 3D model reconstruction of indoor scenes and buildings and their textures based on surfels, buildings and their textures.