In modern times, more and more multimedia applications are implemented on wireless computer networks and used to entertain users through mobile devices. In power-constrained environments such as pocket PCs, PDAs, and cellular telephones, the large amount of video information transmitted from the server-end to the user-end is often compressed to reduce the power and band width consumption. This thesis introduces an efficient method for the construction of motion panoramas and panoramic videos from streaming video. The technique involves the extraction of motion components from the background mosaic which is generated by a hybrid algorithm that combines both feature-based methods and direct methods. Experimental results show this heuristic approach reduces the size of the video information transmitted and summarizes the entire contents of the motion video for the mobile end users.