Video compression based on selectively refining regions in difference frames

We introduce a computationally simple video compression algorithm based on Y. Chiu and T. Berger's (1999) CU30 algorithm, suitable for video streaming and telephony. The authors concentrate on being able to identify and select regions in the difference frame domain in order that only the regions deemed important for quality and rate are chosen and encoded. A binary shape coder is designed to describe the shape of the regions selected. We propose a method of evaluating any binary shape coder by simulating shapes using a two dimensional random field generator, calculating the field's entropy rate, and comparing it to the rate of the encoder.