【摘要】 本文提出了一种基于色度不一致先验理论的雾区检测算法,利用雾的色度特征及雾区的连续性分布特点准确识别和分割雾区,快速给出雾的大致区域及浓度分布情况,利于提升去雾算法的准确度和鲁棒性,对有雾图像的质量评估具有重要意义。结果表明,我们的算法具有很好的检测结果。更多还原

【Abstract】 In this paper,we propose a novel haze-detection method based on a hue disparity prior. It can recognize and segment haze areas according to haze hue and the continuity distribution of haze areas. Our method can estimate the distribution and concentration of haze quickly,which is conducive to improving the accuracy and robustness of dehaze algorithms. It is also of great significance for the quality evaluation of haze images. The results demonstrate that our algorithm is effective and efficient in haze detection.更多还原