Method and system for interpolation in color images

Classical linear signal processing techniques, when applied to color demosaic, tend to over smooth the color signal. This results in noticeable color artifacts along the edges and sharp features. Such techniques, like color filter array (CFA), which each light sensor is covered by one kind of filter, are quite commonly employed in digital cameras for information collection of an image scene, in order to lower costs. A new CFA method which converts multicolor images into full color images by interpolating the missing color pixels with the given color pixels is developed. Each pixel will sample the intensity of just one of the trichromatic color, and the missing color at each pixel can be inferred from the neighboring pixels with a reasonable degree of accuracy. With the color channels supporting the edges and the edges supporting the interpolation of missing colors, it achieves demosaic full color image with better perceptual quality. The smoothing artifact thus produced is much less than some of the known algorithms, while maintaining a computational complexity that is compatible with other fast demosaic algorithms. The application of this new method will allow each pixel of an image to carry the minimal pieces of information while producing full color images with better subjective visual quality