Fast Algorithms for Compositing Multi-Way Compressed Video

Maojun Zhang, Xiaofeng Hu, Bing Yang, Xishu Ku

Abstract: This paper presents fast algorithms for compositing two or more videos which have been compressed using the JPEG (Joint Picture Experts Group) compression standard. Their performance is improved over traditional methods by processing compressed video data in the frequency domain, avoiding the DCT (discrete cosine transform) and IDCT (inverse DCT). Since the algorithms are built on a new theoretical basis, the computational speedup and memory usage efficiency are further improved over existing fast algorithms.