光电子技术, 2013, 33 (1): 27, 网络出版: 2014-01-16
结合图像分割的暗原色先验去雾算法
Image Dehazing Algorithm Combining Dark Channel Prior with Segmentation
图像去雾 暗原色先验规律 基于图论的分割 介质透射率 image dehazing dark channel prior graph-cut segmentation transmission map
摘要
基于暗原色先验规律的图像去雾算法在进行暗原色通道计算时,用固定区域会造成介质透射率估算的不合理,从而导致最终去雾效果细节不清晰和在较远景深场景模糊的现象。采用基于图论的分割算法来确定取暗原色通道的区域,能根据景深的变化自适应地改变区域大小,避免了固定区域求取暗原色的带来的介质透射率估计不合理问题。通过实验分析和验证,证实了该算法能有效地改善雾天图像的退化现象和提高图像清晰度。
Abstract
The classical method based on dark channel prior only uses fixed patch size to calculate the dark channel. This operation makes the estimated transmission map inaccurate and the dehazed image details unsharp with fuzziness in deep scene depth. The graph-cut segmentation algorithm is adopted to confirm the patch size of dark channel and the area size can be changed according to the variation of depth of field adaptively. The inaccuracy of fixed patch size is thus reduced with more accurate estimated transmission map gained. Experimental results show that haze image degradation and the image sharpness are all improved effectively with this algorithm.
许丽红, 王敬东, 邱玉娇, 王子瑞. 结合图像分割的暗原色先验去雾算法[J]. 光电子技术, 2013, 33(1): 27. Xu Lihong, Wang Jingdong, Qiu Yujiao, Wang Zirui. Image Dehazing Algorithm Combining Dark Channel Prior with Segmentation[J]. Optoelectronic Technology, 2013, 33(1): 27.