Real-Time 3D Rendering Based on Multiple Cameras and Point Cloud

Abstract

3D model construction technic is very popular in recent decades. The movie "Avatar" is a milestone of this research, the surrounding, characters and alien creatures in Avatar are so lifelike that viewers would be impressed and shocked by the impact of virtual reality. The video game World of War craft and Diablo are another successful instance about 3D model. Kinect is a 3D sensor produced by Microsoft, and is widely used in many research fields, such as Computer Vision, Computer Graphics-Ketc. With Kinect, the task of efficiently acquiring 3D data would be possible. In other words, the distances between all the positions in Kinect vision and Kinect can be captured and reliable. According to these data, Kinect is able to divide the image into foreground (human) and background quickly and easily draw the geography of human. So this paper would integrate the traditional model building technic and much research about Kinect, and then propose a real-time 3D point cloud displaying method by multiple Kinect.