Research On Image Demosaicing Algorithms

DEMOSAICING algorithms are one of the most important technologies in digital camera product nowadays. Quality of images directly depends on performance of algorithm. If the algorithm is poor performance, photos will appear on the color mosaic obvious, especially in color mutation region, such as object and background edges. Therefore, research on effective demosaicing algorithms has important application value.In this paper, some classic algorithms are firstly reviewed, and their superiority and deficiency are evaluated. Then an improved DEMOSAICING algorithm is proposed for the Bayer CFA based on directional filtering. A directional mark is defined for all the red and blue pixels, then the sum of all directional mark values are computed in a 5×5 neighboring window, to decide direction of edge. The direction of edge was decided more accurately by this method Besides taking algorithmic advantages of several high-effect interpolation algorithms and using directional filtering interpolation, adaptive median filtering is also introduced eliminate aliasing which happens near the edge and to keep the detail in smooth part.In order to assess the performance of the proposed methods, the several test pictures are interpolated by several algorithms in the paper. The fidelity of the reconstructed image is evaluated by computing the peak signal-to-noise ratio and Euclidean distance in CIELAB color space between the original and the reconstructed image in smooth and edge regions respectively. The paper shows the results of quantitative and visual comparison for test images and they verify the effectiveness of the proposed algorithm.