基于目标一维纹理特征的介质辨识方法  

Medium Recognition Method Based on One-dimensional Texture Feature of Target

作　　者：耿琳珊 陈遵田[1] GENG Linshan;CHEN Zuntian(Xi’an Institute of Electromechanical Information Technology,Xi’an 710065,China)

机构地区：[1]西安机电信息技术研究所,陕西西安710065

出　　处：《探测与控制学报》2021年第6期62-67,共6页Journal of Detection & Control

摘　　要：针对光电引信探测装置对地面等实体目标和烟尘气溶胶介质分类的问题,提出基于目标一维纹理特征的介质辨识方法。该方法利用灰度共生矩阵(GLCM)对目标介质一维纹理进行多特征提取,在经过主成分分析法(PCA)降维处理后,利用引信光电探测装置在微距条件下获取的纹理信号特点,构造了K近邻(KNN)分类器,实现光电引信对实体目标和烟尘干扰的分类。仿真验证结果表明,针对目标表面一维强度特征辨识提出的介质辨识技术可有效区分典型的地面介质和烟尘等干扰,并且处理速度快。Aiming at the classification of solid objects such as ground and soot aerosol media by photoelectric fuze detection device,a medium identification method based on one-dimensional texture features was proposed.The method used gray level co-occurrence matrix(GLCM)to extract texture feature.After target media dimension principal component been analysised(PCA),photoelectric detection device using fuze signal characteristics under the condition of macro to obtain texture,the K-nearest neighbor(KNN)classifier was constructs,the goal of photoelectric fuze of entity and the classification of the smoke interference was realize.The simulation results showed that the proposed medium identification technology could effectively distinguish the typical ground medium and the dust and other interference,and the processing speed was fast.

关 键 词：光电引信 灰度共生矩阵 主成分分析 K近邻分类器 

分 类 号：TJ43[兵器科学与技术—火炮、自动武器与弹药工程]