光学 精密工程, 2015, 23 (5): 1466, 网络出版: 2015-06-11   

融合变分偏微分方程的单幅彩色图像去雾

Single color image dehazing using variational partial differential equation
作者单位
空军工程大学 航空航天工程学院 通信导航教研室, 陕西 西安 710038
摘要
对单幅雾霾图像的图像处理问题进行了分析。鉴于现有方法不能很好恢复场景深度变化较大的图像, 本文融合大气衰减模型与变分偏微分方程, 提出了一种单幅彩色图像去雾算法。该算法利用数学形态学中的中值集算子构造局部白平衡处理, 精确估计大气光参数。设计了一种全新的平滑性度量范数, 尝试从全变分的角度建立目标图像的变分能量模型, 继而通过偏微分方程将其转化为欧拉-拉格朗日方程。最后, 利用交替半二次型算法求解欧拉-拉格朗日方程, 以提升算法流程的整体运行速度, 使其可维持在105 ms左右。。以图像熵和平均梯度为图像质量评价指标进行了仿真实验, 结果表明, 本文算法使评价指标提升达60%, 而对照组则均保持在15%至30%左右。该算法对于场景深度变化较大的图像的局部区域恢复效果明显, 可以满足应用要求。
Abstract
Imaging processing for single color hazing images was researched. Since the current methods could not recover the kind of images with the scene depth changing fiercely, a single color image dehazing algorithm was introduced on the basis of the atmosphere attenuate model and the variational partial differential equation. In this method, the median set operator in morphology was used to construct local white balance operator and to estimate the atmospherical optical parameters precisely. Then, a novel smoothness measure norm was designed to build up a variational energy model of the target image based on the total variation theory. In addition, the model was converted from partial differential equation into the Euler-Lagrange equation. Finally, the alternate semi-quadratic algorithm was used to solve the Euler-Lagrange equation, by which the operation speed of the algorithm was improved to be at 105 ms. The means of image entropy and average gradient were taken as the evaluation indexes, and simulation results show that proposed method triggers an increase by 60% in operation performance while other control groups keep the improvement in 15% to 30%. This method improves the local region obviously and reaches the application requirement.

周理, 毕笃彦, 何林远. 融合变分偏微分方程的单幅彩色图像去雾[J]. 光学 精密工程, 2015, 23(5): 1466. ZHOU Li, BI Du-yan, HE Lin-yuan. Single color image dehazing using variational partial differential equation[J]. Optics and Precision Engineering, 2015, 23(5): 1466.

本文已被 3 篇论文引用
被引统计数据来源于中国光学期刊网
引用该论文: TXT   |   EndNote

相关论文

加载中...

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!