Error analysis and suppresion in visual signal processing

Abstract

Error suppression is important for visual signal processing in producing high-quality images and videos. Due to various sources of errors, different suppression techniques are required to meet specific application requirements. The main objective of error analysis is to investigate the principle of error generation, and then incorporate suitable suppression techniques into the design of individual processing steps as well as whole processing chain. An alternative methodology is to analyze the distinct characteristics of the resultant errors, and devise effective removal methods accordingly. This thesis addresses error analysis and suppression in three applications-error suppression in color filter array(CFA)demosaicking, error analy-sis in digital still camera(DSC)image processing chain, and error detection and removal for composite video frames.