TABLE OF CONTENTS It is easy to come up with simple, intuitive methods for compressing images. They are inefficient and are described here for the sake of completeness.
Subsampling is perhaps the simplest way to compress an image. One approach to subsampling is simply to delete some of the pixels. The encoder may, for example, ignore every other row and every other column of the image, and write the remaining pixels (which constitute 25% of the image) on the compressed stream. The decoder inputs the compressed data and uses each pixel to generate four identical pixels of the reconstructed image. This, of course, involves the loss of much image detail and is rarely acceptable. Notice that the compression ratio is known in advance.
A slight improvement is obtained when the encoder calculates the average of each block of four pixels and writes this average on the compressed stream. No pixel is totally deleted, but the method is still primitive, because a good lossy image compression method should lose only data to which the eye is not sensitive.
Better results (but worse compression) are obtained when the color representation of the image is changed from the original (normally RGB) to luminance and chrominance. The encoder subsamples the two chrominance components of a pixel but not its luminance component. Assuming that each component uses the same number of bits, the two chrominance components use 2/3 of the image size. Subsampling them reduces this to 25% of 2/3, or 1/6. The...
