FPGA-based fast rain removal system using orientation-adaptive non-local mean filter
I-An Lin, Trong‐Yen Lee, Chih‐Ming Chen, Shuyu Liu · 2017
Advanced driving assistance systems (ADAS) have developed in widespread use for traffic safety, they must include mechanisms functioning in bad weather conditions. However, the rain reduces visibility and degrade the effectiveness of computer vision algorithm when they are vision-based methods. We propose an orientation-adaptive non-local mean (OA-NLM) filter algorithm to further improve rain removal performance, removing rain streaks in images by high-performance denoising algorithm of NLM and improving computational cost using orientation-adaptive and computing module (CPM). The proposed method is implemented on a Xilinx FPGA system to improve the image processing speed. Experimental results show that the proposed method reduces 94.21% on execution time of image processing with hardware-software design.