Indoor reconstruction method and system based on hyperspectral, true color image and point cloud complementarity

Abstract

Indoor reconstruction method and system based on hyperspectral, true color image and point cloud complementarity, which involves scene adaptation, SVM-based spectrum recognition, point cloud closed-loop correction based on hyperspectral and true color image, distance statistics, greedy triangle and other methods . It solves the problem of point cloud acquisition equipment being insensitive to materials, light and heat, and fully automatic 3D reconstruction. The steps of the present invention are: 1. Obtaining initial data; 2. Identifying the missing position of the point cloud according to the hyperspectral data; 3. Preliminary correction of the three-dimensional point cloud information; 4. Correcting the repair result according to the true color map; 5. Point cloud Filtering, stitching automation, point cloud tessellation and texture mapping. The present invention uses hyperspectral and true-color image information to repair point clouds, which makes up for the lack of hardware and improves the quality of reconstruction. Through adaptive processing and full automation, the stability and practicability of 3D indoor reconstruction are improved. Get promoted, the application prospect is more and more broad.

Description

technical field [0001] The present invention is applied to indoor three-dimensional reconstruction in the field of virtual reality, and specifically uses hyperspectral information, three-dimensional point cloud information and true-color RGB image information to complete high-quality reconstruction of indoor three-dimensional scenes. Background technique [0002] 3D reconstruction technology is a popular research target in computer graphics and computer vision, ranging from the drawing of maps to the reconstruction of grain structures. It can be convenient for scientific research, and the application prospect is more and more broad. [0003] In recent years, in terms of indoor 3D reconstruction, there are binocular 3D reconstruction based on 2D images, real-time positioning and map construction (Simultaneous Localization And Mapping, SLAM) based on 3D depth information, etc. There are many products that collect depth information in the market, but because flat objects with ...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to propose an indoor 3D reconstruction technology that integrates hyperspectral, 3D point cloud and true color image data, which solves the problem of missing point clouds caused by the insensitivity of light sources to materials, lighting conditions and heat during traditional 3D reconstruction. , proposed a method of using hyperspectral information to fill in missing point clouds, using color image data to correct point clouds, combined with other filtering techniques, and finally improved the quality and stability of 3D reconstruction, and further improved the application level

Images