一种改进的暗通道先验去雾算法研究  

Research on an Improved Dark Channel Prior Defog Algorithm

作　　者：韩亮 张爱萍[1] 田东平 HAN Liang;ZHNAG Ai-ping;TIAN Dong-ping(College of Science,Xi'an University of Architecture and Technology,Xi'an Shanxi 710055,China)

机构地区：[1]西安建筑科技大学理学院,陕西西安710055

出　　处：《计算机仿真》2021年第8期455-459,496,共6页Computer Simulation

基　　金：西安市碑林科技局项目资助(GX1902)。

摘　　要：针对暗通道先验算法在图像的明亮区域去雾时存在色偏严重、色彩失真等问题,提出一种改进的暗通道先验图像去雾算法。先将待去雾图像利用三通道明亮区域分割法获得全局大气光强度值;再通过边界约束条件得到透射率图像,并运用高斯型同态滤波对图像进行平滑处理;最后用色阶补偿原理结合小波变换对透射率图像与原图像进行融合,之后将融合后的图像与去雾后的图像进行多次融合补偿,得到最佳色阶图像,达到复原图像的目的。仿真结果表明,上述方法可以改善明亮区域色彩失真问题,保持图像整体轮廓结构清晰及边缘细节突出,提高图像色彩饱和度及整幅图像可视度。An improved dark channel prior image defogging algorithm is proposed to solve the problem of severe color deviation and color distortion in the bright region of image defogging.Firstly,the global atmospheric light intensity of the image to be defogged was obtained by three-channel bright region segmentation.Secondly,the transmission image was obtained by boundary constraint conditions,and the image was smoothed by Gaussian homomorphic filtering.Finally,the principle of color scale compensation was combined with wavelet transform to fuse the transmittance image and the original image,and then the fused image and the defogging image were fused and compensated for many times to get the best color scale image and achieve the purpose of image restoration.Simulation results show that this method can improve the color distortion in bright areas,keep the overall contour structure clear and the edge details prominent,and improve the color saturation and the visibility of the whole image.

关 键 词：图像去雾 大气散射模型 边界约束 高斯型同态滤波 小波融合 色阶补偿 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]