Data Centers

Two Parallel Strategies for Real-Time Spatial Video Denoising for Multi-Core Processors

Download Now Free registration required

Executive Summary

Video denoising is usually a time consuming process especially for large video files. With the advancement of the processor technology, it is possible to perform video denoising in real-time on multi-core processors. In this paper, the authors study parallel techniques for denoising real-time video on multi-core processor which work on both shared memory model and distributed memory model. They investigate two approaches: a block approach, which assigns a group of threads to each block of video frames; and a distributor approach, which uses one thread to distribute the frame data to each thread. Their experiments focus on the image denoising technique based on the total variation but the approach can be integrated with other image denoising algorithm like Discrete Wavelet Transform (DWT) or diffusion technique.

  • Format: PDF
  • Size: 527.14 KB