Enhanced Image View Synthesis Using Disparity Depth Extraction For Efficient Performance In Stereo Vision Applications

Abd Manap, Nurulfajar and Sulaiman, Riza and Darsono, Abd Majid and Mat Ibrahim, Masrullizam and Hairol , Izzuddin (2017) Enhanced Image View Synthesis Using Disparity Depth Extraction For Efficient Performance In Stereo Vision Applications. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text Enhanced Image View Synthesis Using Disparity Depth Extraction For Efficient Performance In Stereo Vision Applications.pdf - Submitted Version Restricted to Registered users only Download (972kB)

Abstract

Image view synthesis draws significant research interests due to its potentials for developing 3DTV and freeviewpoint video applications.The hardware for multi-camera systems is developing fast and are already being deployed for multimedia,security and industrial applications.However,there are still some issues in terms of processing and computational requirement,due to the sheer amount of data involved.The main objective of this research is to develop a new technique to synthesize inter-view images through disparity depth extraction, which potentially lead to more efficient performance in stereo vision applications.Stereo images will be captured simultaneously and calculated through stereo matching algorithm to create disparity depth maps.The disparity depth map will be used to synthesize the inter-view images based on the depth layer extraction.The 3D information will be extracted from the disparity depth map and used for 3D image and free-viewpoint applications.These processes will reduce the complexity by taking advantage of the multi-view information structures.The depth image layer separation technique will be used to create the proposed new algorithm.These new structures and design are anticipated to improve visual performance of human observers based on PSNR and Structural Similarity Index parameters and provide additional views with fewer cameras arrangement compared to the conventional high volume camera configurations for free-viewpoint video acquisition.The algorithm of this research can potentially be used in the education, robotic,surveillance,entertainment and human-vision aided system.

Actions (login required)

View Item