HLS Accelerated Noise Reduction Approach Using Image Stacking on Xilinx PYNQ
Dimitris Tsiktsiris, Dimitris Ziouzios, Minas Dasygenis · 2019
Eliminating noise from the original signal is still a demanding problem for researchers. There has been a number of published algorithms and each approach has its limitations, advantages, and constraints. Here we present our own contribution on the noise reduction problem by proposing an intermediary step between the image sensor and the post-processing software in the image capturing process. Our accelerated method, compared to other researchers, can be used either as a standalone solution or combined with an existing algorithm to further results improvement. Our solution matches the discrete samples between multiple frames and averages the pixel values. The output image maintains its structural integrity, holds better color accuracy and incurs less noise than others.