颅骨性别鉴定在法医学和面貌复原等领域具有重大研究意义和应用价值.本文以陕西关中地区94例汉族成年人三维颅骨数据为研究对象,利用自主研发的系统软件标定出62个显著特征点.在此基础上提取可测量和非可测量特征,并将非可测量特征进行量化表示;提出使用非线性降维方法和非线性分类器结合的方法对颅骨的性别进行分类.实验结果表明:本文整合的可测量和非可测量特征包含更多的性别判别信息,非线性降维与非线性分类结合的方法可以大幅提高性别判别精度,最高正确率可达到96.36%.
Sex identification based on 3Dskull morphology has great significance and application value in forensic science and in facial reconstruction.Ninety four(94)skull samples from Shaanxi Province were used in the present study.Home-made software and knowledge of forensic science and anatomy were used to calibrate62 salient feature points;Both non-measurable and measurable feature points were extracted,and the nonmeasurable characteristics of the skull were quantified.The combination of nonlinear dimension reduction and nonlinear classifier enabled sex determination.It was found that all fusion characteristic can obtain more gender discrimination information.The method of nonlinear dimension reduction and nonlinear classifier enhanced the accuracy and outperformed most of the state-of-the-art methods.