Research And Application Of 3D Reconstruction Of Point Cloud Stitching Based On RGB-D Camera

Three-dimensional reconstruction is a kind of technology which collect the information from the video,image and various range sensors.It used the tools of statistics,geometry and mathematical optimization to make 3D models of the real world.The photos and other information used in 3D reconstruction are taken form the real world,so this technology will apply in the field of 3D printing,real-time positioning and reconstruction,humam-computer interaction and so on.It is also a bridge which connect the real world and virtual world.Three-dimensional digital technology based on PC and the Internet makes people's understanding of reality return to the original visual three-dimensional realm.RGB-D cameras like Microsoft Kinect2.0 is novel sensing systems.It captures RGB images along with per-pixel depth information.In this paper,we acquire 3D Point cloud models from RGB-D cameras and propose a projected image based matching-points searching method to accelerate ICP iteration in point clouds registration.Under certain condition,this algorithm solved the GPU memory limitations problem.The problem is we can not get a large model in one-time on a normal PC.The experimental results proved the possibility of this algorithm.In latter part of this paper,a depth image palm detection algorithm based on histogram and LBP feature is proposed,As the most direct and natural interaction between humam and the world,gesture interaction will become the trend of future human computer interaction.The premise of gesture interaction is to separate the hand apart form the background and reconstruct 3D hands in the virtual world.The process provides a rich and accurate data for the subsequent work of recognition and 3D reconstruction.This technology can be applied to the agricultural realistic 3D model interaction,virtual reality,augmented reality and 3D model reverse engineering.