自适应加权的总变分正则化图像超分辨率重建

A Super-Resolution Algorithm Based on Adaptive Weighted Total Variation

在Farsiu提出的双边总变分正则化方法中,尺度加权系数和正则化参数为定值,在边缘、纹理区域,重建图像效果不理想。针对这个问题,提出了一种自适应加权正则化函数和正则化参数的重建算法,该方法利用图像局部结构信息控制权函数形状、带宽和正则化参数,使这些参数根据图像局部结构信息自适应地调整。对所提出的算法进行了仿真实验,实验结果表明,与传统的总变分重建方法相比较,该算法能更好地重建图像的纹理细节,重建图像的对比度高。

In bilateral total variation regularization method （BTV） , which was proposed by Farsiu, the scale weight is a constant, so the image reconstruction effect is not ideal for texture and edge region. In order to solve this problem, an adaptive weighted regularization function and regularization parameter algorithm is proposed in this paper. In the proposed algorithm, the local structure information of image is used to control the shape, bandwidth of the weighted function and the regularization parameter. The experimental results show that the proposed algorithm, compared with BTV, can retain the image details better and improve the image contrast.