GPU and cell phone-aided multimedia processing

Computing technology has been evolving rapidly during the past decades. New ideas and inventions are constantly developed to improve usability and processing power of applications. This thesis develops a multimedia processing system that uses GPU and cell phone to improve speed and user experience. The CUDA framework developed by NVIDIA turns the GPU into a manycore coprocessor of the CPU. We show in this thesis that by taking advantage of GPU computing, algorithms such as image encoding and resolution upconversion can be up to five times as fast. We also develop algorithms to use accelerometer-equipped cell phone as a remote controller to improve user interaction. When user performs various actions, input command is sent to the PC via Bluetooth and identified using motion detection algorithms. Some applications of these tools are presented, including image slideshow, multiview video viewer, as well as cell phone aided Google Map application and web browser.