image thresholding otsu method in the discrete blocks of size defined by the user in the matlab code.This depends on the colormap of the image that whether the image is gray scale or RGB.
FOR RGB IMAGE: If the image is RGB then we will first extract the luminance and chrominance components of the image and then we will do chroma down sampling which will compress only the chrominance part that is only the color information.This is because we are much more sensitive to changes in luminance (brightness) than we are to chrominance (color) differences. Because of this, the JPEG format can discard a lot more color information than luminance in the compression process. To facilitate the different compression requirements of the two "channels" of image information, the JPEG file format translates 8-bit RGB data (Red, Green, Blue) into 8-bit YCbCr data (Luminance, Chroma Blue, Chroma Red). Now, with the brightness seperated into a separate data channel, it is much easier to change the compression algorithm used for one channel versus the others. Then we will apply DCT compression algorithm in the blocks of (NXN) to further compress the image.After that we will apply otsu thresholding method to rank each block.The ranking of each block is done on the basis of minimum class variance.Those blocks will get the higher rank whose no. of pixels have value greater intensity value than the minimum class variance. After that quantization of the blocks is done on the basis of ranks.Those blocks will be quantized more who got higher ranks.Then the image is encoded and the compressed image and the rank information is stored in a different matrix.
For Gray Scale Images:
In gray scale Images there is no need to transform the image into ycbcr.As there is only luminous part only so there is no need of chroma sampling also.Rest of the process remains the same for gray scale images.