Performance Analysis of Median Filter Demosaicking Algorithm Using New Extended Bilinear Demosaicking

Abstract

Digital image is an approximation of real-world scene. Imaging sensor being one of the most costly components of a digital camera, most of the digital cameras use a single imaging sensor along with a color filter array (CFA) for acquiring a full-color image. The quality of such a full-color image depends mainly on the effectiveness of the demosaicking algorithm used for interpolating missing color pixels. Median filter demosaicking algorithm is one of the most commonly used demosaicking algorithms for generating full-color image from a CFA image. This paper presents a detailed analysis of traditional bilinear demosaicking algorithm used in median filter demosaicking method and presents a new extended bilinear demosaicking for median filter demosaicking. The proposed modifications in traditional bilinear demosaicking algorithm improve the quality of final reconstructed image. We implemented median filter demosaicking with extended bilinear that increases performance of median filtering-based demosaicking method. Experimental results are reported for all images of Kodak dataset and Laurent Condat (LC) Image dataset, and performance is measured using peak signal-to-noise ratio (PSNR) metric. We also compared the proposed method with current state-of-the-art-related approaches and observed that the proposed method performs better.