Handbook of Image and Video Processing: BTC 5.2 BLOCK TRUNCATION CODING (BTC)

The problem of how one stores and transmits a digital image has been a topic of research for more than 40 years and was initially driven by military applications and NASA. The problem, simply stated, is how does one efficiently represent an image in binary form? This is the image compression problem. It is a special case of the source coding problem addressed by Shannon in his landmark paper [1] on communication systems. What is different about image compression is that techniques have been developed that exploit the unique nature of the image and the observer. These include the spatial nature of the data and of the human visual system. The “efficiency” of the representation depends on two properties of every image compression technique: data rate (in bits/pixel) and distortion in the decompressed image. The date rate is a measure of how much bandwidth one would require to transmit the image or how much space it would take to store the image. Ideally one would like this to be as small as possible. If the decompressed image is exactly the same as the original image, the technique is said to be lossless. Otherwise the technique is lossy and the decompressed image has distortion or coding artifacts in it. Depending on the application, one can often trade distortion for data rate, hence if a user is willing to accept images with more distortion the data rate can often be lower. Statistical and structural methods have been developed for image compression [2], the former being based on the principles of source coding with emphasis on the algebraic structure of the pixels in