一种适用于图像拼接的自适应模板匹配算法

在图像拼接中通常使用模板匹配方法进行图像配准,模板特征选取、基准模板位置与大小选取在很大程度上影响图像配准的准确度和速度。根据人类观察事物的行为特点,该文基于局部边缘密度LED和模糊聚类算法定位出特征物体区域,并在此区域上构建基准模板,在匹配上具有更高的可靠性。在模板匹配中采用加权相似性度量的方法,提高相似性度量的可靠性。据此,提出了一种强壮的自适应模板匹配算法,在自动图像拼接中体现出很强的实用性。

A Robust Template Matching Algorithm for Automatic Image Stitching

Template matching is often used in image registration which is the basic component of image stitching.The feature,location and the size of reference template affect the accuracy and speed of image registration greatly.After analyzing the behaviour of human observation,this paper presents a robust self-adaptive template matching algorithm.In this algorithm,the reference template will be constructed at the position of the feature object area whose locating is based on local edge density and fuzzy clustering algorithm,and to improve the reliability of similarity measure,a weighted similarity measure method is used,so this method has high robust in matching.It is very suitable for applications about auto image stitching.