摘要
针对双边全变差正则化算法边缘区域重建图像效果不理想问题,提出一种基于边缘增强的正则化超分辨率图像重建算法。该算法在构造初始图像时,对样条插值后的图像先进行非局部均值滤波预处理,然后进行拉普拉斯锐化处理;采用L1范数度量数据保真项和正则项,并从自适应的角度确定正则化参数,从而增强算法的稳健性。实验结果表明,与样条插值算法、双边全变差算法相比,该算法能更好地增强重建图像的边缘信息。
A regularized super-resolution image reconstruction algorithm based on edge enhancement is proposed for the problem of non-ideal image reconstruction in edge region in bilateral total variation regularization algorithm.While constructing the original image,the spline interpolated image is firstly smoothed by using non-local means filter,and followed by Laplace sharpening processing. The algorithm adopts L1 norm to measure data fidelity item and regularization item.Moreover,the regularization parameter is adaptively determined to enhance the robustness of the algorithm.Experimental results show that the proposed algorithm can better enhance the edge information of the reconstructed image compared with the spline interpolation algorithm and the bilateral total variation algorithm.
出处
《西安邮电大学学报》
2016年第6期14-19,共6页
Journal of Xi’an University of Posts and Telecommunications
基金
国家自然科学基金资助项目(41504115)
公安部科技强警基础工作专项资助项目(2014GABJC024)
陕西省国际科技合作与交流计划资助项目(2015KW-005)