文档介绍：尺度不变的非局部均值超分辨率算法
李翘楚1,杨撒博雅1,刘家瑛1,郭宗明1
1
北京大学计算机科学技术研究所,北京
100080
摘要:本文提出了一种全新的尺度不变的非局部均值图像超分辨率算法。首先,各帧间的尺度
差异值通过本文提出尺度检测算法进行计算得到。基于检测到的尺度变化,本文将不同尺度的
图像块提取并修正到同一尺度,因此在使用非局部均值方法重建图像时可以从图像块中获取更
多的冗余信息。最终,本文采用基于块匹配的图像重建的方法来重建得到高分辨率帧。相比较
非局部均值超分辨率方法,本文提出的尺度不变的非局部均值方法可以从不同尺度的相邻帧中
找到更多的被非局部均值方法忽略的相似块。实验结果表明无论从从主观效果上还是从客观衡
量标准的比较上,本文提出的方法都表现得比前人提出的非局部均值方法及相关算法更好。
关键词:超分辨率;非局部均值;尺度检测;多帧重建;
中图分类号: TP391
Scale Invariant Non-local Means
Super Resolution
Qiaochu Li1, Saboya Yang1, Jiaying Liu1, Zongming Guo1
1
Institute puter Science and Technology, Peking University, Beijing 100080
Abstract: In this paper, we propose a novel scale invariant multi-frame super resolution
(SR) method. First, the scales of each frame are detected by scale-detector. Based on the
scale changes between adjacent frames, patches are extracted and modiﬁed from diﬀerent
scales into the same scale to obtain more redundant information to reconstruction. Finally, a
reconstruction approach based on patch matching is applied to generate a high resolution
(HR) frame. Compared to original Nonlocal Means SR (NLM SR), the proposed Scale
Invariant NLM ﬁnds more potential similar patches in diﬀerent scales which are easily
neglected in NLM SR. Experimental results demonstrate better performance of the proposed
algorithm in both objective measurement and subjective perception.
Key words: Super-reso