Indoor environment reconstruction method based on monocular camera

Abstract

The invention discloses an indoor environment reconstruction method based on a monocular camera, and the method comprises the steps: shooting indoor photos at different angles and positions through the monocular camera, extracting the characteristics of each image through a Harris corner detection algorithm, and obtaining the characteristic point of each image; performing feature point matching onany two pictures with similar shooting positions to obtain a matched feature point pair set of all the images; due to the fact that mismatching feature point pairs exist in matching, mismatching feature point pairs in the matching feature point pair set are eliminated; through a motion inference structure SFM, reconstructing a sparse point cloud for the photo with the mismatching feature point pair eliminated; carrying out dense point cloud reconstruction on the sparse point cloud, reconstructing all point clouds in the scene, and recovering the indoor scene environment. According to the method, the complexity of a common feature extraction method is reduced through a feature extraction algorithm with higher speed, feature point matching is carried out through a kd tree, sparse three-dimensional point cloud is reconstructed according to a geometric method, and a better reconstruction effect is realized through a dense reconstruction algorithm in the later period.

Description

technical field [0001] The invention relates to the technical field of three-dimensional reconstruction, in particular to a method for indoor environment reconstruction based on a monocular camera. Background technique [0002] People perceive the world through three-dimensional information. From images, we can only understand the two-dimensional information of things, but not their three-dimensional information. At the same time, with the rapid development of virtual reality, augmented reality, and autonomous driving technologies, the technology of reconstructing 2D images to 3D perception is becoming more and more important, such as superimposing virtual 3D models in the real environment and automatically acquiring video during autonomous driving. The captured depth information and so on are inseparable from two-dimensional to three-dimensional reconstruction technology. [0003] Existing 3D reconstruction techniques can be roughly divided into two categories, 3D reconstr...

AI Technical Summary

Problems solved by technology

However, the limitation of the reconstruction based on the geometric method is that its calculation amount is very large, the effect of real-time reconstruction is very general, and due to the complexity of the indoor environment, the reconstruction effect is often not ideal

Images