指纹(计算)
线性判别分析
背景(考古学)
法医学
拉曼光谱
模式识别(心理学)
偏最小二乘回归
人工智能
化学计量学
犯罪现场
法医人类学
分析化学(期刊)
计算机科学
化学
机器学习
色谱法
心理学
生物
光学
地理
考古
物理
犯罪学
遗传学
作者
Marco Antonio de Souza,Alexandre Silva Santos,S.W. da Silva,Jez Willian Batista Braga,Marcelo Henrique Sousa
标识
DOI:10.1016/j.forc.2021.100395
摘要
From the late 90 s until recently, some forensic research has been dedicated to the development of analytical techniques to explore the chemical components present in fingerprints, in order to find other information besides authorship. Raman spectroscopy is a technique of nondestructive analysis of a wide variety of forensic evidence, including fingerprints, at the crime scene. In this context, the aim of this work is to explore Raman spectroscopy and the supervised methods, Partial Least Squares and Support Vector Machine for Discriminant Analysis (PLSDA and SVMDA, respectively), as means to determine sex based on fingerprints obtained from male and female donors and submitted to different conditions (dark and light). Considering a period up to seven days from the collection of the fingerprint, the results showed correct discrimination rates ranging from approximately 80–93%.
科研通智能强力驱动
Strongly Powered by AbleSci AI