一种基于组合矩和小波变换的目标匹配算法  

A Fast Algorithm for Object Matching Based on Combined-moment Invariant and Wavelet Transform

作　　者：赵超君[1] 王敬东[1] 李鹏[1] 

机构地区：[1]南京航空航天大学自动化学院,南京210016

出　　处：《光电子技术》2009年第1期18-22,共5页Optoelectronic Technology

摘　　要：不变矩方法是一种经典的特征提取方法。文中针对纹理细节信息较少的特定目标,首先在七个不变矩的基础上构建了组合矩。该组合矩具有旋转、平移、比例不变性和良好的类间可分离性,大大降低了算法的复杂性、提高了抗干扰能力。为了进一步提高匹配速度,引入多分辨率小波分解,利用低频系数来计算图像的组合矩特征量进行模板匹配。实验表明,对于细节信息较少的目标,该方法是一种具有较高实时性和鲁棒性的目标匹配算法。Moment invariant is a classical character extracting method.In view of the specific objects with few detail information,the combined-moment invariants based on Hu 7 moment invariants are constructed,which have the rotation,translation,proportion invariability and the good separability.The method also reduces the algorithm complexity and improves the antinoise ability.In order to increase the matching speed further,the wavelet transform is introduced into template matching,which uses the low frequency inform...

关 键 词：不变矩 组合矩 小波变换 目标匹配 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]