Image Gap Interpolation for Color Images Using Discrete Cosine Transform

Image Gap Restoration in color image is a developing field in image processing. The objective of my work is to restore the missing blocks in a digital color image. The method adopted here is a blend of interpolation and multiscale transformation. One of the best framework available to restore image gaps is multiscale transformation. It transforms an input image into a set of downscaled samples until the block size reduces to one. The popular methods used for image transformation are DCT (discrete cosine transform) and DWT (discrete wavelet transform). I am following DCT method in my project work. Two types of interpolation methods are used in my project work. In the first methods, the value of the missing pixel is obtained by taking the edge-weighted mean of the available neighboring pixel. The second method is an edge based interpolation. In this method, first, we find out the main edges of an image and its direction using any edge detection method like canny, sobel etc. the missing pixel value is calculated by using the available neighboring pixel along the same direction. The proposed algorithm gives improved PSNR (peak signal to noise ratio).