摘要
提出一种简单、通用的基于正则化技术的自适应MAP超分辨率重建算法。与以往算法不同,该方法引入了局部空间自适应正则化参数,弥补了传统算法对图像自身的局部特性缺乏考虑的不足。算法通过迭代的方式,利用中间重建结果不断对正则化参数进行更新,并最终得到重建图像。实验结果表明,该方法可以根据不同图像序列的特点以及图像的局部灰度特性,自适应地确定相应的正则化参数,并找到最优解,有效地保护了高分辨率图像的细节信息。
A simple and universal super-resolution reconstruction algorithm of adaptive MAP based on regularized technique is proposed. Differing from the conventional ones, the new algorithm introduces local spatial adaptive regularized parameter, which overcomes the shortcoming that the local characteristic of the image itself is not considered in conventional algorithms. The regularized parameter keeps updating at each iteration step with the partially reconstructed results to obtain final reconstructed image via iterative method. The results of the experiments indicate that the proposed algorithm can determine the regularized parameter automatically and adaptively according to the characteristics of different image sequence as well as local grey character of the image, and find the optimal result, so the detail information in high resolution image is protected effectively.
出处
《计算机应用与软件》
CSCD
2009年第12期238-240,共3页
Computer Applications and Software
关键词
超分辨率
图像重建
空间自适应
正则化
Super-resolution Image reconstruction Spatial adaptive Regularized
