Hand-held 3D Dense Map Construction System Based on Improved ICP Algorithm.

In view of the shortcomings of low-dimensional modeling, difficult to move and expensive, the 3D reconstruction system based on the improved ICP (Iterative Closest Point) algorithm is proposed. Firstly, the Kinect sensor is employed to collect color images and depth images, and the Kinect pose is calculated by tracking and acquiring image feature points, which effectively improves the camera's error in obtaining depth images and image noise, and realizes local map construction. Based on the ORB-SLAM (Simultaneous localization and mapping based on ORiented brief feature recognition) system, the key frames are extracted to obtain more image pixels. The improved ICP algorithm is presented to construct a three-dimensional dense map, and a good three-dimensional model of indoor scene is obtained. The experimental results show that the proposed high-resolution three-dimensional reconstruction system with large scenes has good stability, accuracy and robustness, and has great application value.