雾天降质图像的去雾复原新算法  

A New Algorithm for Restoration of Fog-degraded Images

作　　者：张鑫[1] 麻金继[1,2] 吴浩 

机构地区：[1]安徽师范大学,安徽芜湖241003 [2]安徽师范大学自然灾害过程及防治研究重点实验室,安徽芜湖241003

出　　处：《遥感信息》2015年第5期42-47,共6页Remote Sensing Information

基　　金：国家自然科学基金(41271377)

摘　　要：针对雾天可见光图像对比度低,成像效果差的情况,提出了一种图像去雾新算法。该算法在暗原色先验知识的基础上,结合引导滤波函数,解决了雾天图像场景透射率不精确的问题。算法结果与暗原色先验方法得到的结果相比,复原效果明显,不仅还原了雾天场景的轮廓和颜色信息,而且对不同材质的物体表现更佳,使得去雾后图像的可视性增强,更加贴近真实场景。同时,对于浓雾天气条件下获取的图像,去雾能力显著,计算效率更高,大大降低了去雾算法的复杂程度。For the fog visible images with low contrast and poor imaging conditions,we propose a new algorithm for the foggy images.The algorithm based on a priori knowledge of Dark Channel and combining with Guided Filter Function greatly solves the problem of inaccurate scene transmittance of the images due to the heavy haze.Compared to the results of Dark Channel Priori,the effect of the restoration of this algorithm is obvious,which not only restores the contour and color information of foggy scenes,but also different materials of the objects can be better performed.The results of de-hazing enhance the visibility of the image and make it closer to the real scene.Meanwhile,under heavy fog weather conditions,the ability to defog is significant.The algorithm with high computational efficiency can greatly reduce the complexity of de-hazing.

关 键 词：图像去雾 物理模型 透射率 引导滤波 图像增强 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]