Title :
CFA Demosaicking with Improved Colour Edge Preservation
Author :
Randhawa, Sharmil ; Li, J. S Jimmy
Author_Institution :
Sch. of Inf. & Eng., Flinders Univ., Adelaide, SA
Abstract :
Colour edges are sharp boundaries between two distinct colours. Most existing demosaicking methods depend on assumptions which will be invalid at edges. In this paper, we apply a method that avoids interpolation across an edge by extrapolating estimates in four different directions. In order to improve colour edge preservation, we propose a novel classifier to select the best estimate in conjunction with its surrounding pixels. It has been shown that our method outperforms visually and quantitatively with image quality measures, when compared with other existing methods.
Keywords :
filtering theory; image resolution; image segmentation; interpolation; CFA demosaicking; colour edge preservation; colour filter array; image quality measures; Australia; Digital cameras; Equations; Extrapolation; Humans; Image edge detection; Image quality; Informatics; Interpolation; Strips;
Conference_Titel :
TENCON 2005 2005 IEEE Region 10
Conference_Location :
Melbourne, Qld.
Print_ISBN :
0-7803-9311-2
Electronic_ISBN :
0-7803-9312-0
DOI :
10.1109/TENCON.2005.301070