摘要
针对雾天可见光图像降质现象,提出一种简单、有效的自动去雾算法。算法充分利用了近红外透雾能力强,可见光图像色彩亮度好的特点。首先利用暗原色先验估计出雾浓度分布,然后根据此分布计算出归一化暗原色图像亮度分布系数,并对其进行双边滤波处理,最后利用该系数将近红外与可见光图像进行融合。实验结果表明,该算法能有效地改善雾天图像质量,且具有广泛的应用前景。
A novel method for automatic haze removal was proposed. The algorithm made the best use of advantages of both NIR and visible light image. First,the haze density distribution was estimated by using dark channel prior. Based on this,the normalized luminance distribution coefficient of dark channel image was calculated and treated via bilateral filtering,and used to fuse near-infrared( NIR) and visible light image together. Experiments showed that the algorithm can effectively improve the quality of the degraded image and has broad application prospects.
出处
《四川大学学报（工程科学版）》
CSCD
北大核心
2013年第S2期155-159,共5页
Journal of Sichuan University (Engineering Science Edition)
基金
国家"973"重点基础研究发展规划资助项目(2009CB320803)
国家"863"高技术研究发展计划资助项目(2012AA011804)
关键词
近红外
图像融合
图像去雾
滤波
near-infrared
image fusion
image dehazing
filtering
