支持向量机方法在线条痕迹检验中的应用  

Application of support vector machines in examination of worn striation mark

作　　者：景畅[1] 王炳成[2] 任朝晖[3] 

机构地区：[1]广东警官学院技术系,广东广州510232 [2]深圳大学法学院,广东深圳518060 [3]东北大学机械工程与自动化学院,辽宁沈阳110004

出　　处：《辽宁工程技术大学学报（自然科学版）》2006年第2期264-267,共4页Journal of Liaoning Technical University (Natural Science)

基　　金：广东省科研基金资助项目(2003C32401)

摘　　要：应用支持向量机的基本原理和信息融合思想,经对两种不同类型枪支所发射同一型号子弹上磨损线条状痕迹数据提取特征后,用特征小波包维数和磨损痕迹中坡膛痕迹尺度参数作为样本数据的因子分量,建立了较为理想的SVM分类模型,以此SVM分类模型实现了对磨损线条状痕迹的分类识别。研究表明:采用参数融合要比使用单一参数更有利于分类识别,并能提高分类识别率。This paper, with the basic principle of support vector machine, theory and information fusing idea, extracts the characteristics of worn marks data of the same type of bulletin shot by two different guns, uses trait wavelet package dimension and size index of sloping-chamber marks among worn marks as factor measure of sample data, and establishes an ideal classification model, and then realizes the classification recognition of worn striation mark by SVM classification model. This research indicates that using index fusing is more effective than using only one of the indexes to classification recognition, and its classification recognition rate can be improved.

关 键 词：支持向量 信息融合 线条状痕迹 分类识别 

分 类 号：TJ27[兵器科学与技术—武器系统与运用工程]