基于不变性模糊特征融合的红外目标识别方法  

Infrared Target Recognition Method Based on Invariant Fuzzy Feature Fusion

作　　者：杜婕婕[1] 刘峥[1] 

机构地区：[1]西安电子科技大学雷达信号处理重点实验室,陕西西安710071

出　　处：《制导与引信》2005年第4期56-60,共5页Guidance & Fuze

基　　金：国家部委预研基金资助课题(编号:51407020104DZ0127)

摘　　要：研究了将平移、缩放与旋转变化(TSRI)的不变性模糊特征信息融合技术用于红外目标图像识别分类的方法,通过提取红外目标图像奇异值与不变矩的平移、缩放与旋转变化(TSRI)的不变性特征,结合模糊特征融合的模式识别方法,并应用模糊判决准则进行分类。实验结果表明,基于红外图像的TSRI不变性多特征信息融合的目标识别方法比不具有TSRI不变性多特征信息融合的识别方法具有更好的稳定性、准确性和可靠性,能够有效地提高红外成像末制导识别系统的精确度,并增强了系统的容错性能。Invariant fuzzy feature information fusion technique based on the invariance of translation, scale and rotation in-plane（TSRI） is used to recognize and classify the infrared image targets in this paper. First extracts the singular value feature and the invariant matrix feature of TSRL from the infrared target image, then uses the fuzzy recognition on basis of the fuzzy feature fusion. Some decision strategies are used to recognize and classify the infrared images. The results of the experiment show that the TSRI invariant multi-feature information fusion method based on the infrared imaging is more stable, more accurate and more reliable than the general multi-feature fusion method without TSRI, and can efficiently improve the accuracy and the ability of fault tolerance in the system of the infrared imaging terminal guidance recognition.

关 键 词：不变性 模糊识别 红外目标 特征选择 

分 类 号：TN215[电子电信—物理电子学] TP391.4[自动化与计算机技术—计算机应用技术]