视频实时去雾的硬件优化设计

Hardware Optimization Design of Video Real-time Defogging

为满足视频去雾实时性的要求,提出了基于硬件优化的视频实时去雾架构。该架构采用增量更新和定常数转换等方式,对大气光值和环境光值等计算模块进行硬件优化设计,从而提高片内资源的重复利用率,实现计算精度的控制。经过MATLAB和Modelsim联合仿真验证后,应用于FPGA芯片上。结果表明:所提架构的数据处理速度高达100.8 Mpixel/s,对pixel的视频图像去雾处理可以以60 fps的速度稳定运行。

In order to meet the real-time requirements of video defogging,this paper proposed a real-time video defogging architecture based on hardware optimization.The architecture used incremental update and constant conversion to optimize the hardware design of calculation modules such as atmospheric light value and ambient light value,thereby improving the reuse rate of on-chip resources and realizing the control of calculation accuracy.After the joint simulation and verification of MATLAB and Modelsim,it was applied to FPGA chip.The result shows that the data processing speed of the proposed architecture is as high as 100.8 Mpixel/s,and the defogging processing of the video image of pixels can run stably at a speed of 60 fps.