Vision is the most important source of information for human beings. With advances in electronic and information technology, the importance of Image acquisition and Display technology in conveying information to our vision system is ever prominent. Image acquisition and display technology have greatly extend our vision, enabling us to see things that can’t be seen with naked eye, such as distant galaxy, micro world and the ability to see though our own body. However, the physical world that we are living in is a 3D world, while the traditional 2D image acquisition and display technology have limited our perception capability. The emerging 3D Imaging and 3D Display technology break the limits of traditional 2D imaging and display technology, greatly enhance our ability in information acquisition and comprehension. 3D image acquisition and display technology are two major branches of 3D imaging technology and they are an inseparable whole. 3D images acquired with 3D image acquisition are the major source of content for 3D display, while the acquired 3D images can’t be accurately presented without 3D display. There already exist several different types of 3D image acquisition and 3D display technologies. Based on existing works, we developed several 3D imaging systems and 3D display systems and investigated the possibility of applying 3D imaging technology in some clinical applications. Several different attempts to 3D image acquisition are presented in this work, which include an active 3D image acquisition system, a passive 3D reconstruction system for endoscope images and compressive light field imaging. The active 3D image acquisition system consists of a projector and a pair of cameras. Using a GPU based sub-pixel matching algorithm, it can perform real-time 3D image acquisition with high accuracy. The passive 3D reconstruction system uses shape from motion and bundle adjustment to recover scene structure, and then uses patch based multi view stereopsis to generate dense 3D point cloud. Surface mesh can be generated from the point cloud with Poisson surface reconstruction technique. The light field imaging system starts with sampling the light field randomly, by using l1 minimization and an over-complete light field dictionary, the light field can be restored. We also developed several different 3D display systems which include a multiview 3D light field display and a computational multi-layer LCD 3D display. The multi-view 3D light field display c...
修改评论