基于改进加权引导滤波的煤矿井下图像除雾算法研究

A Defogging Algorithm for Coal Mine Underground Images Based on Improved Weighted Guided Filtering

针对暗通道先验理论在煤矿井下尘雾图像处理中出现的图像失真、颜色偏暗、细节丢失等问题,提出了一种基于亮暗通道先验理论的煤矿井下图像除雾算法。首先,融合暗、亮双通道先验知识,重构局部大气光;其次,分析双边滤波的滤波特性,建立基于局部像素差值的改进加权引导滤波算法;最后,利用暗通道大气光和原始图像的像素差值自适应调节图像的整体亮度。试验结果表明,改进加权引导滤波算法能够有效除去煤矿井下图像中的尘雾影响,与基于引导滤波的He算法和Retinex-MSR算法相比,本算法拥有较高的平均梯度和标准方差,复原后的图像清晰度更高,单张图像处理时间均低于1 s,比Retinex-MSR算法运行时间缩短了近1/2。

Aiming at the problems of image distortion,dark color and detail loss in coal mine dust and fog images processing based on dark channel prior theory,a defogging algorithm for coal mine underground images based on bright and dark channel prior theory was proposed.Firstly,the prior knowledge of dark and bright channels was fused to reconstruct the local atmospheric light.Secondly,the filtering characteristics of bilateral filtering were analyzed,and an improved weighted guided filtering algorithm based on local pixel difference was established.Finally,the pixel difference between the dark channel atmospheric light and the original images was used to adaptively adjust the overall brightness of the images.The experimental results show that the improved weighted guided filtering algorithm can effectively remove the influence of dust and fog in the coal mine underground images.Compared with the He algorithm based on guided filtering and Retinex-MSR algorithm,this algorithm has higher average gradient and standard variance,and the restored images have higher clarity.The processing time of a single image is less than 1 s,which is nearly 1/2 shorter than the running time of Retinex-MSR algorithm.