Processing of visual binary data in the compressed domain using generalised finite transducers

The paper investigates compressed domain processing of binary visual data, namely binary images and binary shape masks. Generalised finite transducers are used for manipulation of binary images and shape masks. The experimental results are presented that show the gain in speed when processing visual data in the compressed form. Moreover, no quality loss occurs when visual data is processed in this way. Finally, the potential applications of this technique are suggested and shortcomings are discussed.