广义全变差正则化图像去噪模型研究

RESEARCH ON GENERALIZED TOTAL VARIATION REGULARIZATION MODEL FOR IMAGE DENOISING

本文研究了全变差正则化模型在图像去噪过程中易产生阶梯效应的问题,依据图像的局部结构特利用联合高斯滤波器和边缘检测算子的方法,构建了广义全变差正则化图像去噪模型,获得了在消除噪声的同时能够保留图像边缘细节和纹理信息的结果.实验结果表明,广义全变差正则化模型在平滑噪声的同时能够保留图像的边缘轮廓等细节信息,得到的复原图像在峰值信噪比、平均结构相似度和主观视觉效果方面均有所提高.

To deal with the problem of staircase effects during noise removal, a modified generalized total variation denoising model is proposed in this paper. To guarantee the satisfactory denoising performance, the generalized parameter is calculated adaptively by combining the Gaussian filter and edge detection filters based on locM image features. The proposed model could keep a good balance between noise reduction and image details preservation. The experimental results have demonstrated the superior denoising performance of the proposed total variation model in terms of the peak signal to noise ratio, mean structural similarity and subjective visual effect.