lossy compression is acceptable for images if further analysis or use of the images will tolerate missing data.

The following are common lossy compression techniques [2]:

• Joint Photographic Experts Group (JPEG)

• Moving Picture Experts Group (MPEG)

The following are representative lossless compression techniques [2]:

• Lempel-Ziv and Welch (LZW) algorithm

Binary images such as black-and-white text and graphics are good candidates for lossless compression. Some color and gray-scale images that contain series of successive identical pixels may also be compressed significantly with lossless compression. Pixels of color images commonly vary in chromatics, since adjacent pixels may have different color values. Frequent change in color requires storing the bits for a large number of pixels because there is little repetition. The same applies to gray-scale images where a large number of shades are present. In these cases, lossless compression techniques may not produce a reduction in size sufficient to be practical. Generally, when lossless compression techniques do not produce acceptable results, lossy approaches are used in a way that minimizes the effects of the loss. The payoff of compression becomes generally questionable with animated images and full motion color video.

All current image compression techniques utilize one or more of the blocks shown in Fig 2.

If the image data is in red-green-blue (RGB) format it may be changed to hue saturation value (YUV). This process is applied to color images to change the relationship between pixels and produce an image that is more appropriate for human perception. MPEG uses this format and reduces Y and

V by 2, thereby obtaining some image compression at this stage.

JPEG and MPEG use the discrete cosine transform (DCT), which is given for an N xN image f(x, y) by

0 0

Post a comment