改进的暗通道先验快速图像去雾算法  

Improved fast image defogging algorithm based on dark channel prior

作　　者：张江鑫[1] 周佳波 孟利民[1] 

机构地区：[1]浙江工业大学信息工程学院,杭州310023

出　　处：《计算机应用》2017年第8期2324-2328,共5页journal of Computer Applications

基　　金：国家自然科学基金资助项目(61372087)~~

摘　　要：针对暗通道去雾算法复杂度过高的问题,提出一种基于暗通道的快速去雾算法。首先对图像分块处理,提高求取暗通道值的速度;接着采用线性插值算法平滑处理,消除块状效应;然后根据暗通道先验求得透射率;最终利用大气散射模型得到清晰无雾图像。实验结果证明,在保证良好去雾效果的前提下,相比原暗通道算法,所提算法复杂度被有效降低,去雾所需时间也降为原来的1/10左右,达到准实时。Due to high complexity of dark channel prior defogging algorithm, a fast defogging algorithm based on dark channel prior was proposed. Firstly, the computing of the dark channel map was accelerated by blocking the image. Secondly,the block effect was eliminated by using the linear interpolation algorithm to smoothing. Then, the transmission map was acquired by dark channel prior. Finally, a clear and haze-free image would be gotten from the atmospheric scattering model.The experimental results show that the defogging effect of the proposed algorithm can be as good as the original defogging algorithm but the complexity can be reduced effectively. The proposed algorithm needs time about one-tenth of the original algorithm and reaches the requirement of near real-time.

关 键 词：去雾算法 暗通道 导向滤波 分块 线性插值 

分 类 号：TP391.413[自动化与计算机技术—计算机应用技术]