摘要
图像匹配技术广泛应用于各种图像处理任务,如图像拼接、机器视觉等。通常匹配算法的精确度只能达到像素级别,但在很多图像处理任务如超分辨率重建中需要亚像素精度的图像配准。提出了一种基于相位相关的亚像素图像配准算法。根据两幅离散数字图像的相位相关矩阵中的最大值以及其附近若干点可以拟合估计出实际的峰值位置,进而实现两幅图像的亚像素运动估计。提出的算法针对热像仪采集的红外图像进行匹配实验,实验结果表明该算法精度相比通常的亚像素匹配算法较高,且具有更好的实用性。
Image registration technology is widely used in various image processing tasks such as image mosaic, machine vi- sion etc. Generally, many registration methods can only achieve accuracy of pixel order. However, many image processing tasks including super-resolution reconstruction need the registration accuracy of sub-pixel level. This paper mainly proposes a sub-pixel image registration method based on phase correlation. Based on the maximum value and its neighborhood of phase correlation matrix between the two discrete images, the location of the real peak value of phase correlation matrix can be ob- tained by fitting the several points, and then the sub-pixel motion estimation can be implemented. The proposed algorithm is validated by implementing a series experiments, the experimental results show that the proposed approach performs better than the conventional methods.
出处
《光学与光电技术》
2013年第6期15-18,共4页
Optics & Optoelectronic Technology
关键词
图像匹配
亚像素
相位相关
插值
image registration
sub-pixel
phase correlation
interpolation
