Streaming Thin Client Compression

Thin client compression (TCC) achieves the best compression efficiency for sequences of synthetic images. This paper presents a streaming version of TCC (STCC) that almost fully retains the excellent compression efficiency of the original algorithm. When sending images over low-bandwidth networks, STCC dramatically reduces end-to-end latency by pipelining rows and overlapping the compression, transmission and decompression stages. We describe a streaming boundary tracing algorithm and tree-structured codebook that allow inherently two-dimensional pattern matching and substitution methods to be implemented efficiently in a streaming manner. We found that streaming and pipelining reduce end-to-end latency by up to 40.6% over the original TCC method.