Manipulating lossless video in the compressed domain

Abstract

A compressed-domain transformation is one that operates directly on the compressed format, rather than requiring conversion to an uncompressed format prior to processing. Performing operations in the compressed domain offers large speedups, as it reduces the volume of data processed and avoids the overhead of re-compression. While previous researchers have focused on compressed-domain techniques for lossy data formats, there are few techniques that apply to lossless formats. In this paper, we present a general technique for transforming lossless data as compressed with the sliding-window Lempel Ziv algorithm (LZ77). We focus on applications in video editing, where our technique supports color adjustment, video compositing, and other operations directly on the Apple Animation format (a variant of LZ77). We implemented a subset of our technique as an automatic program transformation. Using the StreamIt language, users write a program to operate on uncompressed data, and our compiler transforms the program to operate on compressed data. Experiments show that the technique offers speedups roughly proportional to the compression factor. For our benchmark suite of 12 videos in Apple Animation format, speedups range from 1.1x to 471x, with a median of 15x.