基于FPGA的视频实时去雾算法及其硬件实现

单幅图像去雾技术虽然已经取得较大的进展,但是算法较为复杂,运行时间较长.为了实现视频实时去雾,以硬件实现为目的,对暗通道先验算法进行改进,降低其时间复杂度.提出了一种暗通道图优化方法,保留了图像的边缘信息,消除了光晕效应,省去了透射率细化的复杂操作;提出了适应于硬件实现的大气光值估计和调节及透射率补偿方法,解决了视频帧间闪烁及天空等明亮区域的色彩失真问题.基于现场可编程门阵列(FPGA)对所提出算法进行了硬件实现.结果表明,该算法可以实时处理帧速为60 f/s、分辨率为1 920×1 080的视频图像,相比传统去雾算法速度更快,去雾质量更高.

FPGA-based Video Real-time Dehazing Algorithm and Hardware Implementation

Single image dehazing technology has made great progress, but the algorithm is complicated and the running time is long. For the purpose of hardware implementation, the dark channel priori algorithm is improved to reduce its time complexity and realize real-time dehazing. A dark channel map optimization method is proposed, which retains the edge information of the image, and eliminates the halo effect and the complicated operation of transmittance refinement. The method for estimating and adjusting atmospheric light, and a transmittance compensation method suitable for hardware implementation are proposed, which solves the problem of flicker between video frames and color distortion in bright areas such as the sky. The hardware implementation of the proposed algorithm is based on FPGA, and the results show that it can process video images with the resolution of 1920×1080 at the frame rate of 60f/s in real time, which is much better than that of the traditional dehazing algorithm.