摘要
针对混合噪声,结合加权稀疏与变分,提出了新颖的去噪模型。首先,进行PCA训练自适应字典,再结合非局部相似性,利用噪声的特性进行加权编码。最后,结合变分正则项,再利用对偶方法求出恢复后的图像。仿真实验表明,该算法不仅提高了图像的峰值信噪比,而且更好地保留图像的重要特征,提高图像的视觉效果。
Aiming at the mixed noise, a novel denoising model based on weighted sparse and variational is proposed in this paper. First, PCA method is used to train adaptive dictionary, and then, combined the non-local similarity with the characteristics of noise, it can get the weighted coding. Finally, the restored image is obtained by using the dual method. Experimental results showed that this algorithm can not only improve image＇ s peak signal to noise ratio （ PSNR）, but also preserve the important fea- tures, which result in improving the visual quality of the image.
出处
《电视技术》
北大核心
2016年第10期33-36,49,共5页
Video Engineering
基金
国家自然科学基金项目(61362021)
广西自然科学基金项目(2013GXNSFDA019030
2012GXNSFBA053014
2014GXNSFDA118036)
广西高校图像图形智能处理重点实验室项目(GIIP201408
GIIP201503)
关键词
混合噪声
加权编码
非局部相似
变分
对偶方法
mixed noise
weighted coding
non-local similarity
variational
dual method
