Fast embedded video compression using cache-based zerotree processing

An embedded coding algorithm creates a compressed bit stream which can be truncated to produce reduced resolution versions of the original image. This property allows such algorithms to precisely achieve fixed bit rates, reducing or eliminating the need for rate control in video transmission applications. Unfortunately, a large amount of scratch memory is required for embedded coding, making it difficult for a cache-based processor to continuously operate at full speed. We address this problem in two steps. First, we partition the wavelet coefficients into separate groups and code them sequentially using the embedded zerotree wavelet algorithm. Next, we use information gathered in the first step to optimize the bit allocations for each partition. The changes to the allocation are transmitted to the decoder and applied to the next video frame.