Algorithms and architectures for real time image compression using a feature visibility criterion

Reviews several low complexity algorithms for image compression that use a feature visibility threshold to remove detail that is not visible under normal viewing conditions. A quadtree decomposition of the image is used to facilitate an orderly examination of progressively larger image areas. These methods are combined with efficient lossless coding of the resulting “pruned” quadtree image. Variants on the basic quadtree algorithm and the extension of these methods to color images are also discussed. This paper emphasizes architectural issues in the design of hardware for the real time compression and decompression of high bandwidth images using these algorithms