Software Acceleration of Video Noise Filtering and its integration into real-time video applications
The project is mainly in the domain of achieving real-time computational speed of methods to remove noise from video signals (for example, those taken by a professional cinema camera). Specifically, in this project, we propose first to improve the speed of current technology that we have developed in previous MITACS projects, in order to make it commercially valuable and second to integrate this new real-time technology into video applications that require noise-free inputs in order for them to have high performance output. The architecture of personal computers allows us to use several processing units simultaneously and this lead to speed up significantly but the program code should be well managed to use all of these resources together. Since currently our code does not have this feature and processing speed is slower than desired, by modifying the program we expect to meet the acceptable speed. We are targeting two important applications: video compression and face recognition. Early investigation shows that we will noticeably improve their performance by either removing noise from their input or estimating that noise.