基于FPGA的透射率快速估计去雾方法

A Dehazing Method by Fast Estimating Transmittance Based on FPGA

在雾、霾等天气条件下,由光学相机所获取的原始景物会出现严重的图像退化现象,比如对比度及能见度降低、图像模糊等情况。目前在图像去雾领域,暗通道先验算法对解决上述问题具有良好的效果,但是考虑到该算法的计算复杂度以及实时性,无法将其直接移植到FPGA内。因此在暗通道先验算法基础上,首先用一种大气透射率快速估计方法来降低计算量,同时改善传统算法出现的Halo现象,然后采用自动色阶方法对去雾后图像进行对比度拉伸,改善去雾效果。实验结果表明,所提算法不仅满足实时去雾的要求,而且有效提高了图像去雾的能力。

Under the weather conditions such as fog the original scene acquired by the optical camera will have serious image degradation for example the contrast and visibility of the image are reduced and the image is blurred.At present in the field of image dehazing the dark-channel priori algorithm has a good effect in solving the above problems.But in view of the computational complexity and real-time performance of the algorithm it cannot be directly transplanted into the FPGA.Therefore based on the dark-channel priori algorithm this paper first uses a fast estimation method for atmospheric transmittance to reduce the calculation amount and at the same time weakens the Halo phenomenon of the traditional algorithm.Then the auto level method is used to perform contrast stretching on the dehazed image for improving the dehazing effects.Experimental results show that the proposed algorithm not only meets the requirements of real-time dehazing but also effectively improves the ability of image dehazing.