基于Hausdorff距离的图像配准快速算法  

Fast Algorithm for Image Registration Based on Hausdorff Distance

作　　者：杨通钰[1] 彭国华[1] 

机构地区：[1]西北工业大学理学院,西安710129

出　　处：《计算机工程》2011年第12期193-195,198,共4页Computer Engineering

基　　金：国家自然科学基金资助项目(60672135)

摘　　要：在图像配准过程中,传统Hausdorff距离算法的计算量较大。针对该问题,提出一种基于Hausdorff距离的图像配准快速算法。将参考图像和待配准图像进行边缘检测,在待配准图像上任意选取一个模板,通过设定一个变化的阈值对Hausdorff距离算法进行改进,以减少不必要点的计算,实现快速匹配,并根据匹配数据,对图像进行尺度变换及旋转操作,使2幅图像能在空间上配准。实验结果表明,与传统的配准算法相比,该算法的计算复杂度较低。In the image registration process, the computational complexity is the major problem of the traditional Hausdorff distance algorithm. This paper proposes an image registration fast algorithm based on Hausdorff distance. Edge examination to the reference image and the image is registration, and then chooses a template, sets a variable threshold to improve the Hausdorff distance algorithm so as to reduce the amount of unnecessary to achieve rapid match. According to the matched data, it carries on the scale transformation and the rotation operation to the image, enables two images registration in the space. Experimental result proves that, comparing with the traditional matching algorithm, the computational complexity reduced greatly.

关 键 词：模板匹配 图像配准 边缘检测 二值化 Hausdorff距离算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]