摘要
针对多视频源图像的拼接问题,提出基于多尺度分析的图像配准及融合方法。在配准方面,弥补传统的基于Harris角点的配准方法对尺度变化不稳定的缺点,使其具有多尺度特性,提取的角点更加稳定精确。在融合方面,采用基于多尺度多分辨率分析的图像融合方法,克服传统空间域图像融合算法在频域上的不足,具有在空间域和频率域的局部化能力,并提供人眼视觉对其比较敏感的强对比度信息。实验结果表明,相比较传统方法,该方法所提取角点更精确、融合效果更好。
An image registration and fusion method based on multi-scale analysis is proposed, which focuses on multi-source video mosaic. In terms of registration, the method compensates for the lack of traditional registration method, which is unstable on scale changes, and makes the comer detection much more stable and accurate. In terms of fusion, in order to overcome the shortcomings of traditional fusion algorithm in frequency domain, multi-resolution image fusion method based on multi-scale analysis is adopted, which not only has localized capacity in both space domain and frequency domain, but also provides more sensitive and strong degree information for human visual. The feasibility of the algorithm is verified by experimental simulation. The result shows the extracted comer is more accurate and fusion effect is more ideal compared with traditional methods.
出处
《计算机工程》
CAS
CSCD
2013年第9期237-239,244,共4页
Computer Engineering
基金
国家自然科学基金资助项目"基于压缩感知的高效CMOS图像传感器研究"(61076024)
国家"863"计划基金资助项目"基于可重构处理器的视觉信息处理应用系统设计"(2012AA012705)
关键词
多尺度分析
视频拼接
多源视频
角点检测
特征匹配
图像融合
multi-scale analysis
video mosaic
multi-source video
comer detection
feature matching
image fusion
