Realization of 3D Reconstruction Algorithm Based on 2D Video

At present, how to realize 3D reconstruction from 2D video is a research hotspot. However but the reconstruction accuracy for areas with less or no texture is low. Thus, for the scene reconstruction with less texture, this paper proposes a 3D reconstruction method based on the depth map. The first is to extract the relevant feature points from the two-dimensional image frame. During this process, Harris-Sift algorithm is used to extract features of the image frame, and Kanade-Lucas-Tomasi (KLT) tracking algorithm is used to tracke and matche feature points. In this way, all adjacent image frames are reconstructed by 3D point cloud, and the recondtruction result is optimized through the bundle adjustment algorithm. Furthermore, the depth map is calculated, and the depth map of each frame in the video is obtained by using the planar scanning algorithm. Finally, the three-dimensional reconstruction of the object is realized by fusing the depth map, and point cloud conversion. As shown by the experimental results, the real scene is accurately restored by the relevant algorithm provided in this paper, and the problem of reconstruction of areas with less texture is solved.