A New Approach to Decoding and Compositing Motion- Compensated DCT-Based Images

Multi-point network video services require compositing several video sources into a single displayed video stream. Compositing video directly in the compressed domain can save computations by processing less data and avoiding the conversion process back and forth between the compressed and uncompressed data formats. Earlier work has demonstrated the computational speedup of compositing DCT-compressed video directly in the DCT domain, compared to the straightforward spatial-domain approach. Typical compositing operations include overlapping, scaling, translation, filtering, etc. In this paper, we propose a new decoding algorithm for MC-DCT compressed video which converts MC-DCT compressed video to the DCT domain and enables video compositing in the DCT compressed domain. Computational complexity is analyzed. The idea of network compositing and its impacts on multimedia network implementations are also discussed. CD ROM Title and Abstract decompression, and increased latency for transmitting video to end users. Our experiments show that accumulation errors depends on the compositing operations performed, for example, more severe degradation for scaling, ignorable degradation for translation and overlapping. We have simulated the DCT-domain compositing algorithms for different compositing scenarios and quality levels. For the compositing scenario shown in figure 6, our proposed DCT-domain algorithm can save the computations by 24%, compared to the straightforward spatial-domain approach. If the compression is without the MC algorithm (like JPEG), the reduction of computations is about 73%. The reconstructed images do not show clear subjective difference between our compressed-domain approach and the spatial-domain approach. (The PSNR is 42.8 dB vs. 43.2 dB.) The non-zero DCT coefficient percentage varies between 5%~22%, and the non-zero motion vector percentage ranges between 12%~90% in our simulations. The speedup can be increased if we raise the compression ratio by sacrificing some image quality. We design a new decoding algorithm for the MC-DCT based video, which performs inverse MC before inverse DCT. This algorithm can be applied in compositing compressed video within the network, which may take multiple compressed video sources and combine them into a single compressed output stream. The proposed algorithm converts all MC-DCT compressed video into the DCT domain and performs compositing in the DCT domain. This DCT-domain approach can reduce the required computations with a speedup factor depending on the compression ratio and the non-zero motion vector percentage. However, dropping some least-significant DCT coefficients maybe necessary for the worst case of high-motion video in real-time implementations. Some issues of networked video compositing are also discussed. Another direct application of …