An efficient measurement of image processing and fusion algorithms for color medical image demosaicing

Abstract

Color demosaicing or color interpolation is an important image newlineprocessing technique as it determines the resulting digital image quality of the newlinecamera. The ideal demosaicing algorithm would minimize the complexity newlinewhile maximizing quality. However, in reality, trade-offs must be between newlinecomplexity and quality. Most of the digital cameras capture the scene with a newlinesingle sensor, which be coated with a Color Filter Array (CFA) in order to newlinereduce the cost and size of the camera. Many types of RGB CFA patterns are newlineavailable. In this thesis, Bayer CFA patterns are used which contains the newlineprimary colors (Red, Blue and Green) in the particular format. This Bayer newlineCFA pattern captures these colors separately. The green color band is newlinepreferred twice than that of the other two primary colors. Because the newlinesampling density of green band will be higher where the other two bands are newlinelower. This captured image is the CFA image or mosaic image. This mosaic newlineimage contains incomplete color samples and needs to interpolate first by newlineusing the demosaicing algorithms. Then, color interpolation is utilized, to newlineobtain the color image by estimating the missing components. Many color newlinedifferences and residual based interpolation techniques are analyzed. This newlinemethod gives the better PSNR values and visual quality. In the residual newlinemethod, color artifacts arise since the bilinear interpolation is used in the newlinegreen channel reconstruction. In order to overcome these artifacts, a novel newlinedemosaicing algorithm is implemented in this thesis. newline newline