Color image compression using multi scale mean shift segmentation

In the colorization based compression methods, the representative pixels, which represents the color information of the input image are used to reconstruct the color image at the decoder. The selection of the representative pixels from the input color image is an important issue. In all the previous methods, the representative pixels are selected in an iterative manner which is more time consuming and reduces the quality of the image In this method the selection of the representative pixels are done in a better way. Here, it is guaranteed that the selected representative pixels are optimal set which reduces the error between the original image and the reconstructed image. A colorization matrix is also constructed here which colorizes the luminance of the image using different scales. Thereby we will get a image having good quality with reduced size. The experimental results shows that this method surmounts the conventional methods as well as the comparison has been done with the JPEG and JPEG2000 both in terms of file size and the quality of the reconstructed image.