摘要
针对传统Retinex变分模型采用相同的权重容易引起虚假痕迹的缺陷,通过引入差分特征值作为边缘指示算子,构造了一种具有空间自适应调节能力的Retinex变分校正模型。该模型能够利用影像空间域信息来控制变分校正模型在不同像素点的约束强度,在边缘区域施加较小的正则化约束保持影像的边缘特征;而在平坦区域施加较大的正则化约束。同时根据反射分量的物理性质,在变分校正模型中对其施加均值逼近灰度中值约束防止局部曝光过度。采用分裂Bregman迭代法实现对该变分校正模型的最优化求解,利用模拟影像和真实影像进行实验,并与传统方法进行比较,结果表明,该方法能够消除影像灰度不均匀现象,同时大幅提高计算效率。
This paper proposed a spatially adaptive Retinex variational model by using the edge indicator called difference eigenvalue to overcome the disadvantage of the tradition methods that were easy to cause false traces with the same weight.The proposed model could automatically balance the regularized strength among different spatial property regions in the image.In the edges,it enforced weak regularized strength to preserve detail,and in the homogeneous areas,it enforced strong regularized strength to eliminate the uneven intensity.It employed the gray world assumption to constrain the reflectance,and used a fast computation method based on the split Bregman iteration to solve the proposed model.Simulated and real experimental results show the effectiveness of the enhanced image quality and the efficiency of the minimization.
作者
左芝勇
Zuo Zhiyong(Southwest China Institute of Electronic Technology,Chengdu 610036,China)
出处
《计算机应用研究》
CSCD
北大核心
2018年第5期1582-1585,共4页
Application Research of Computers
