一种基于特征点匹配的图像快速拼接方法  

A FAST IMAGE STITCHING METHOD BASED ON FEATURE POINTS MATCHING

作　　者：张麟华[1] 高腾[1] 

机构地区：[1]太原工业学院计算机工程系山西太原030008

出　　处：《计算机应用与软件》2013年第6期272-275,共4页Computer Applications and Software

摘　　要：实现一种基于特征点匹配的图像拼接方法。将处理流程细分为特征点检测、特征点匹配、无效匹配的剔除、透视成像四个阶段,并针对四个阶段选取了适合的经典算法予以实现。其中对具有左右关系且基本平行的图像特征点匹配算法进行了改进,在匹配效果基本不变的前提下,缩短了匹配时间。经实验证明,算法改进后效率提升明显。We achieve an image stitching method which is based on feature points matching. The processing flow is finely divided into four stages： the feature point detection, the feature point matching, the elimination of invalid matching, and the fluoroscopic imaging. Moreover, proper classic algorithms for these four stages are selected to implement them. Among them the image feature points matching algorithm with left-right relationship and being basically parallel is improved, this shortens the matching time without the change in matching effect. Proved by the experiment, the efficiency of the algorithm improves significantly.

关 键 词：特征点匹配 图像拼接 快速 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]