基于多特征融合的图像匹配模式  

Image matching based on multi-features fusion

作　　者：彭真明[1] 张启衡[2] 魏宇星[2] 张覃平[2] 

机构地区：[1]电子科技大学光电信息学院,四川成都610054 [2]中国科学院光电技术研究所,四川成都610209

出　　处：《强激光与粒子束》2004年第3期281-285,共5页High Power Laser and Particle Beams

基　　金：国家863计划项目资助课题

摘　　要：　常规图像匹配模式主要利用了像素的灰度信息和形状信息,而弱小目标检测与跟踪过程中,这两种信息都缺乏明显特征,难于满足高精度、稳定跟踪的要求。提出一种新的匹配模式,即从图像数据里提取包括灰度、形状在内的多种特征信息。寻找一种简单有效的信息融合手段,进而获取一种综合特征,利用"综合特征"进行相似度量来确定目标的最佳定位。仿真计算结果表明,该方法是可行和有效的。The conventional methods of image matching are based on gray and shape of image. These two features can't describe image data well especially in tracking the faint and small targets. It's difficult to track the target with high-precision and steady. A new algorithm of image matching was proposed. It extracts characters including gray and shape from image data and gains an integrated character by using the information fusion algorithm with simplicity and effect. Then calculates the position coordinate of target using integrated characteristics by correlation degree. Simulation shows that the method is feasible and effective.

关 键 词：图像处理 多特征融合 K—L变换 图像匹配 弱目标检测 

分 类 号：O413[理学—理论物理] TL56[理学—物理]