摘要
针对单幅低分辨率灰度图像,提出一种基于稀疏表示和字典学习的超分辨率重建算法,通过选择合适的过完备字典,图像块可表示为字典元素的稀疏线性组合。对于输入的低分辨率图像,寻求每一图像块的稀疏表示,利用此表示系数产生高分辨率图像输出。为消除Elad方法重建图像中产生的黑色边缘并提高重建图像的质量,文中在稀疏表示方法的基础上利用反向投影法对其进行改进。仿真实验结果表明,改进算法不仅实现了上述目的,而且在图像信噪比和算法运行效率上都有所提高,从而达到了算法改进的目的。
Present a super-resolution reconstruction approach to single gray image based on sparse representation and dictionary learning. Image patches can be well-represented as a sparse linear combination of dictionary' s elements from an appropriately chosen ~ver-com- plete dictionary. For each patch of the low-resolution input image, seek a sparse representation and then use the coefficients of this repre- sentation to generate the high-resolution output image. In this paper,in order to eliminate the black edge and improve the image' s quali- ty,introduce the back-projection into the Elad' s super-resolution reconstruction. The results of simulation experiment show the method not only achieves the above purpose, but also lead to a marked improvement both in PSNR (Peak Signal to Noise Ratio) and operating efficiency.
出处
《计算机技术与发展》
2013年第9期103-106,共4页
Computer Technology and Development
基金
江苏省自然科学基金(BK2011758)
南京邮电大学攀登计划(NY208050)
关键词
稀疏表示
图像超分辨率
学习字典
稀疏编码
sparse representation
image super-resolution
learned dictionary
sparse coding