3D Reconstruction From Monocular Images Based on Deep Convolutional Networks

Abstract

3D reconstruction from monocular images is an essential part of machine vision. The effective reconstruction of 3D models of objects or scenes has become a research focus in computer vision. In this article, we concentrate on restoring the 3D model through the depth information of a monocular image. First and foremost, we use deep learning to get the depth of a single image. Next, the point cloud is reconstructed from the depth map according to the camera parameters to achieve the purpose of the 3D reconstruction. Compared with traditional methods, our approach is faster and more relaxed, and the reconstructed three-dimensional information of scenes can be recovered efficiently, our reconstructed 3D models also have better authenticity.