基于多重引导滤波的单幅图像去雾算法

Single Image Dehazing Algorithm Based on Multiple Guided Filtering

雾天拍摄到的图像往往存在颜色、细节信息丢失和对比度降低等问题,这对户外视觉任务如目标检测、识别、分割等带来非常大的挑战。提出一种基于多重引导滤波的单幅图像去雾算法。算法首先利用灰度限制Gamma变换获得一幅低亮度去雾基底图像;其次,利用多重引导滤波获得细节图像;再次,将去雾基底图像和细节层图像求和获得低亮度的去雾结果;最后,通过亮度调整提升低亮度去雾图像亮度,得到最终去雾结果。对比实验验证了该算法能够有效去除雾影响提升图像质量。

Images captured in foggy days often face problems such as color loss,detail information loss,contrast reduction,which brings great challenges to outdoor visual tasks such as target detection,recognition,segmentation,etc.A single image dehazing algorithm based on multiple guided filtering is proposed.Firstly,a dehazing base image is obtained by using gray-limited Gamma transform.Secondly,multi-guided filtering is used to obtain detailed images.Thirdly,the dehazing base image and the detail layer image are summed to get the low brightness dehazing result.Finally,the brightness of low-brightness dehazing image is improved by brightness adjustment to get the final result of dehazing.Through comparative experiments,it is verified that the proposed algorithm can effectively remove the influence of haze and improve the image quality.