題目：Computer Vision Techniques for 3DTV
主講人：賴尚宏教授(國立清華大學資訊工程學系)
時間：100年10月26日 (星期三13:30-15:30)
地點：三峽校區社科大樓社1F06教室
摘要：
3DTV has become the trend of the next-generation TV standards. More and more new imaging and display products related to 3DTV have been released to the market. The capture and display of 3D content heavily rely on the 3D computer vision techniques. In this talk, I will present some 3D computer vision techniques related to 3DTV from two perspectives. One is about the generation of the 3D content from 2D images, and the other perspective is the view synthesis for 3D display. For the generation of the 3D content, I will focus on two approaches, i.e. depth map estimation from a single video and the stereo reconstruction. For the view synthesis problem, I will describe a temporally consistent video inpainting algorithm for novel view synthesis. Some experimental results will be shown to demonstrate these 3D computer vision techniques.
