基于随机森林算法的手印年龄分类研究  

A Study of Age Classifi cation by Handprint Based on Random Forest Algorithm

作　　者：赵瑞敏 刘凯 张丽梅[1] 孙鹏[2] 张忠良[1] ZHAO Ruimin;LIU Kai;ZHANG Limei;SUN Peng;ZHANG Zhongliang(College of Forensic Sciences,Criminal Investigation Police University of China,Shenyang 110035,China;College of Public Security Information Technology and Intelligence,Criminal Investigation Police University of China,Shenyang 110854,China)

机构地区：[1]中国刑事警察学院刑事科学技术学院(法庭科学学院),沈阳110035 [2]中国刑事警察学院公安信息技术与情报学院,沈阳110854

出　　处：《刑事技术》2023年第5期449-458,共10页Forensic Science and Technology

基　　金：公安部重点研究计划(2011ZDYJXJXY001)。

摘　　要：利用手印分析年龄是法庭科学领域的重要研究方向之一,但现存的各式分析方法存在主观经验性强、特征利用率低、选取样本量小和分类效果差等问题。针对上述问题,提出了基于随机森林算法的遗留者手印年龄分类模型。首先,收集1600例男性(11~76岁)右手指纹样本,定量化处理指纹中可能随年龄增长发生变化的特征数据;然后将各特征数据与个体年龄之间进行相关性分析以筛选其中与年龄显著相关的特征;最后,分别基于随机森林、ExtraTrees、CatBoost、KNN和BP神经网络算法建立年龄四分类和六分类模型,通过模型比较,评估随机森林算法的适用性,并输出特征变量重要性排序。实验结果表明,随机森林算法在5种分类模型中的分类效果最佳,具体表现为:四分类模型中随机森林的分类精度为0.812,Kappa系数为0.738;六分类模型中随机森林的分类精度为0.738,Kappa系数为0.675。因此,本文提出的基于随机森林算法的手印年龄分类模型能够更加科学准确地判断个体年龄范围,从而为侦查和技术检验提供线索和范围。The fi ngerprint analysis of age is one of the important research directions in the fi eld of forensic science,but there are some problems with various existing analysis methods,such as strong subjective experience,low feature utilization,small sample size and poor classifi cation effect.To solve the above problems,an age classifi cation model based on random forest algorithm was proposed.Firstly,the right-hand fi ngerprint samples of 1600 males(aged 11 to 76)were collected to quantitatively process the feature data that might change with age.Then,the correlation analysis between each feature’s data and individual age was carried out to screen for features that were signifi cantly correlated with age.Finally,based on random forest,ExtraTrees,CatBoost,KNN and BP neural network algorithms,four and six age classifi cation models were established respectively.The applicability of the random forest algorithm was evaluated by model comparison,and the importance ranking of characteristic variables was output.The experimental results show that the random forest algorithm is the best for classifi cation among the fi ve classifi cation models,which are as follows:in the four-classifi cation model,the classifi cation accuracy of random forest is 0.812,and the Kappa coeffi cient is 0.738;in the six-classifi cation model,the classifi cation accuracy of random forest is 0.738,and the Kappa coeffi cient is 0.675.Therefore,the fi ngerprint age classifi cation model based on the random forest algorithm proposed in this paper can judge the age range of individuals more scientifi cally and accurately,so as to provide clues and ranges for investigation and technical inspection.

关 键 词：指纹 年龄 相关性 随机森林 分类模型 

分 类 号：D918.91[政治法律—法学]