Real-time video denoising on multicores and GPUs with Kalman-based and Bilateral filters fusion

Abstract

In the context of video processing, image noise caused by acquisition, transfer and image compression can be attenuated by video denoising algorithms. However, their computational cost must be as low as possible to allow them to be applied to real-time applications. In this paper, we propose stmkf, a real-time video denoising algorithm based on Kalman and Bilateral filters. We evaluate the effectiveness of stmkf using several common videos used in the literature and we compare it to other denoising algorithms using both the PSNR and SSIM metrics. Our experimental results show that stmkf is competitive with other filters, especially for videos that feature stationary backgrounds such as in videoconferencing, video lectures and video surveillance. We also evaluate the performance of our parallel implementations of stmkf for CPUs and GPUs. stmkf achieved a performance improvement of up to \(2.9\times \) on a Intel i7 multicore processor with 4 cores compared to the sequential solution. The results obtained with the GPU version of stmkf on a NVIDIA Tesla K40 showed a performance improvement of up to \(7.6\times \) compared to the Intel i7 multicore processor.