摘要
图像超分辨率重建技术是数字图像领域的一个研究热点,应用广泛。为了使重建的图像能更好地保持边缘细节,采用各向异性高斯核函数作为适用度函数,并将改进的自适应归一化卷积超分辨率重建算法应用于设计的多通道光学成像系统图像。由于各向异性高斯核函数邻域的尺度和方向由提出的自适应结构张量矩阵决定,其能很好地估计图像局部结构的方向和强度。实验仿真结果表明,提出的方法与其他方法相比可以保持边缘细节和提高信噪比,从而改善图像成像质量。
The technique of image super-resolution reconstruction is a focus in the field of digital image, and applied to a wide range of fields. This paper presents an image reconstruction algorithm that combines the characteristics of multi-channel optical imaging system with improved adaptive normalized convolution super-resolution reconstruction algorithm. The experimental results show that, according to the anisotropic Gaussian kernel obtained by constructing the structure tensor matrix, compared with other methods, the super-resolution image reconstruction based on adaptive normalized convolution can greatly improve signal to noise ratio and improve image quality.
出处
《计算机工程与应用》
CSCD
北大核心
2016年第8期191-195,共5页
Computer Engineering and Applications
基金
安徽高校省级科学研究项目(No.KJ2011Z135)
中科院光电技术研究所微细加工光学技术国家重点实验室开放基金
关键词
多通道成像系统
结构张量矩阵
归一化卷积
图像重建
multi-channel imaging system
structure tensor matrix
normalized convolution
image reconstruction
