Research on Hardware Implementation Technology of Dark Channel Priori Dehazing Algorithm Based on PYNQ

This paper performs a complete study on the dark channel priori dehazing algorithm and proposes an effective method of mapping the dark channel acquisition, minimum filter, mean filter, matrix calculation and atmospheric light calculation involved in the algorithm on FPGA. The dark channel priori dehazing algorithm is implemented as a hardware accelerator on the Xilinx PYNQ platform and the experimental results show that the proposed FPGA accelerator achieves 7.285 speedup compared to performing the algorithm on PC (CPU: Ï7-6700H, frequency:2.6GHz, memory: 4GB DDR3). Its frame rate reaches 18.8 FPS.