论文部分内容阅读
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。本文用分裂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，本文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。
Aiming at the defect that the traditional Retinex variational model can easily cause false marks by using the same weight, a Retinex variational correction model with adaptive spatial adaptive ability is constructed by introducing the difference eigenvalue as the edge indication operator. This model can use the image Spatial domain information is used to control the constraint strength of the variational correction model at different pixels, and a smaller regularization constraint is applied in the edge region to maintain the edge feature of the image. However, a larger regularization constraint is imposed on the flat region. At the same time, according to the physical properties of the reflection component, the mean value is approximated to the gray median constraint in the variational correction model to prevent local overexposure. In this paper, the split Bregman iterative method is used to optimize the model of Variational correction. Experiments using simulated images and real images are carried out and compared with the traditional methods. The results show that the proposed method can eliminate the uneven gray level of the image, Improve computational efficiency.