Based On The Color Ccd Image Interpolation Algorithm

This dissertation mainly discusses CFA demosaicing owing to Bayer CCD image. These algorithms belong to a key technology in digital camera product nowadays. In nowadays society, the digital camera is put into use broadly. For considering cost, the manufacturer adopt the single CCD image sensor, whose surface is covered with a color filter array, The CFA limits each sensor pixel just sampling one of the three primary color values. This dissertation discusses Bayer CFA which is most popular CFA at present. For getting the full color image, the other two missing color samples must be interpolated. This process is commonly referred as CFA demosaicing.In this paper, we first review some commonly used algorithms. Then propose two improved demosaicing algorithms for the Bayer CFA. In high frequency area of image, most general demosaicing algorithms have no way to reproduce true color if object width can only have 1～2 pels on CCD, and put up the color pieces having nothing to do with object content, that these color pieces call as colored moire. In the first improved algorithm, missing green samples are first estimated based on general interpolation. The missing red and blue components are then estimated based on the interpolated green plane. This algorithm can effectively reduce the color artifacts, colored moire. The second improved algorithm is one kind of compound interpolation algorithm, it unite algorithmic advantage of some high-effect interpolation algorithms, and can get fine interpolation effect. Three steps carry out this method mark, are respectively: first interpolation owing to border detecting, high frequency replace, vector median filtering. Various interpolations for 24 photographs in Kodak real color image data base have been in progress, to explain algorithmic superior or inferior rank of various interpolations.In order to assess the performance of the proposed methods, the dissertation also discusses image evaluation method. The peak signal-to-noise ratio and Euclidean distance in CIELAB color space are computed between the original and reconstructed image in whole and edge regions respectively. The dissertation gives out the charts of quantitative comparison and visual comparison. Experimentation validates that algorithm is feasible and can reach good effect.