EDGE STRENGTH BASED FUZZIFICATION OF COLOR DEMOSAICKING ALGORITHMS

Abstract

Most digital camera use only a single photo sensor overlaid with a color filter array (CFA) to capture image data. This allows only one of the required color samples to be available at each pixel location and other two color components need to be interpolated. This process of reconstructing the full color image from the incomplete color components at each pixel output from the image sensor is known as demosaicking or color filter array interpolation. Over the past years, many demosaicking algorithms have been introduced in order to optimize the subjective and objective interpolation quality, it becomes difficult to implement them in digital cameras due to their limited computing capacity, available processing time, and hardware size. An edge strength based fuzzification of demosaicking algorithms is proposed in this thesis in which edge strength information from the raw image data is fuzzified and effectively utilized to improve the interpolation quality of current demosaicking algorithms. We have used five image datasets including the kodak lossless true color image suit to test our approach over three performance measures, MSE, PSNR and computation complexity. Experimental results confirm the effectiveness of our approach when compared to other algorithms.