摘要
针对单一目标特征量提取与识别存在的不足,提出了不变矩与红外目标几何形态特征组成的联合特征量,并选取欧氏距离作为目标分类识别的判别准则。实验结果表明,联合不变矩作为特征量满足尺度不变性,且有很强的独立性与可分性。在目标识别分类的应用中,其效果优于不变矩特征识别。
To overcome the shortage of extracting characteristics and recognizing target on one side, the combined characteristics vector is presented within invariant moments and geometry shape. Furthermore, target recognition is realized by using the distance as the judgment.Experimental results indicate that application of the combine invariant moments as the characteristics and the distance as the judgment can meet the invariant requirement,which has better recognition performance than traditional invariant moments on the using of target recognition.
出处
《沈阳理工大学学报》
CAS
2016年第2期10-13,29,共5页
Journal of Shenyang Ligong University
基金
国家863计划资助项目(2009AA04Z167)
关键词
红外图像
目标识别
不变矩
联合特征向量
欧式距离
infrared image
target recognition
invariant moments
combined characteristics
euclidean distance
作者简介
张旭艳（1990-），女，硕士研究生
通讯作者：华宇宁（1965-），女，教授，研究方向：信号与信息处理、虚拟仪器。
