叶懋, 黄品高, 吕洋, 唐宁, 刘建伟. 基于FPGA的雾霾视频图像实时复原系统研究[J]. 应用光学, 2019, 40(5): 812-817. DOI: 10.5768/JAO201940.0502004
引用本文: 叶懋, 黄品高, 吕洋, 唐宁, 刘建伟. 基于FPGA的雾霾视频图像实时复原系统研究[J]. 应用光学, 2019, 40(5): 812-817. DOI: 10.5768/JAO201940.0502004
YE Mao, HUANG Pingao, LYU Yang, TANG Ning, LIU Jianwei. Research and implementation of real-time fog and haze video image restoration system based on FPGA[J]. Journal of Applied Optics, 2019, 40(5): 812-817. DOI: 10.5768/JAO201940.0502004
Citation: YE Mao, HUANG Pingao, LYU Yang, TANG Ning, LIU Jianwei. Research and implementation of real-time fog and haze video image restoration system based on FPGA[J]. Journal of Applied Optics, 2019, 40(5): 812-817. DOI: 10.5768/JAO201940.0502004

基于FPGA的雾霾视频图像实时复原系统研究

Research and implementation of real-time fog and haze video image restoration system based on FPGA

  • 摘要: 针对雾、霾等恶劣气象条件下, 视频监控等户外视频设备对视频图像实时复原的需要, 提出一种快速图像复原算法的FPGA实现方法, 通过适当降低算法复杂度达到了降低时间开销和硬件实现复杂度的目的。通过将一帧视频分解成3×3小窗口进行运算, 以窗口中心像素的流水计算替代一帧视频像素计算, 使每个像素在3个时钟周期内完成全部运算, 确保了系统能够实时复原, 降低了复原模块对硬件的开销。实验证明:FPGA处理结果和MATLAB处理的结果相符, 可以在保证除雾效果的前提下, 实时处理分辨率为640×480视频并以29帧/s速度流畅显示。

     

    Abstract: Aiming at the need of video surveillance and other outdoor video equipment for real-time restoration of video image under severe weather conditions such as fog and fog, an field-programmable gate array(FPGA) implementation method for fast image restoration algorithm was proposed. By appropriately reducing the complexity of the algorithm, the purpose of reducing time cost and hardware implementation complexity could be achieved.In this algorithm, a frame of video is decomposed into 3×3 small windows for operation, and a frame of video pixel calculation is replaced by a pipeline calculation of the center pixel of the window, so that each pixel's all operations are processed in 3 clock cycles, ensuring that the system can be restored in real time, and the hardware cost of the recovery module can be reduced.Experimental results show that the results of FPGA processing coincide with the results of MATLAB processing, the real-time processing resolution is 640×480, and the video can be displayed smoothly at 29 frames /s under the premise of ensuring the defogging effect.

     

/

返回文章
返回