CCD IMAGE DEMOSAICING USING LOCALIZED CORRELATIONS

摘要

A new approach to image interpolation using spatialrelationships between adjacent pixels is introduced. In itsfirst stage, the localized statistical relationships arestudied based on the sparse version of the image. In thesecond stage, the governing rules of the image are used tobuild an interpolated version. The proposed interpolationmethod is suitable for color single-CCD images fordemosaicing purposes. The correlation rule is studiedfirst for each color component separately, then differenceimages (modified hues) are built to eliminate the colorcorrelation, leading to a smoother reconstructed signal.Since in Bayer pattern not all the color components areequally represented, the algorithm deals with the majorgreen component differently from the red and blue, usingthe green as a basis for the whole image reconstruction.Further statistical tools are added to the algorithm toimprove the visual results. We compare our method topresently available demosaicing techniques for single-CCD color imaging with the major emphasis on reducingghost colors and unreal edges. Our conclusion is that theproposed method can significantly improve interpolationand demosaicing tasks in image processing.