APPLICATIONS OF PARTITIONED ITERATED FUNCTION SYSTEMS IN IMAGE AND VIDEO COMPRESSION

Iterated function systems (IFS) have been used to compress image data. Because of difficulty in finding IFS in natural images, a technique b ased on partitioned IFS (PIFS) has been proposed for image compression . In this technique, an image to be compressed is divided into nonover lapping blocks. For each block an affine transformation is found in th e image. This set of affine transformations (called PIFS) corresponds to a unique image. In the simplest case, images are partitioned into f ixed size blocks. In this paper, we investigate image and video compre ssion techniques using variable block sizes based on the quadtree part ition. One property of images generated using PIFS is scalability: the y have fine detail in any scale. We exploit this property to reduce re quired compression time and improve compression performance. There are large amounts of temporal redundancy between fames of a video sequenc e. We describe a method to remove temporal redundancies effectively us ing a quadtree partitioning technique. We have implemented the above s chemes to compress image and video sequences and will report our exper imental results. (C) 1996 Academic Press, Inc.