Designing and Realization of 3D Video Capture and Transfer System Based on Kinect

Abstract

This paper presents the design and implementation of a real-time 3D video capture and transfer system based on Microsoft Kinect. On the collecting layer, multiple views of texture images and depth images are acquired by 3 Kinect cameras, which are placed as a parallel array. Hole filling and noise filtering techniques are used to fill holes and smooth noise in depth image on the processing layer. On the coding layer, video stream is processed using stereo video coding or multi-view video plus depth coding methods to deal with the huge amounts of data, and then transported in network on the transport layer. At last, media stream is decoded at the decoder layer, and shown on the stereoscopic display. Experimental results show that the proposed system presented in this paper can be well effective to fill the holes and reduce the noise for depth image. Stereo video coding and multi-view video plus depth coding method are also included in this system. This system can be applied to 3D video conferences, 3D entertainment and 3D security monitoring, etc.