分析测试学报2025，Vol.44Issue(11)：2339-2345,7.DOI:10.12452/j.fxcsxb.25020568
基于高光谱成像技术和随机森林的原子印油种类识别
Study on Identification of Atomic Oil Types Based on Hyperspectral Imaging Technology and Random Forest Algorithm
摘要
Abstract
The identification of stamp pad ink types is a critical component in the field of forensic document examination,holding significant practical value for the analysis of seal impressions.In this study,hyperspectral imaging technology was employed to acquire spectral data from multiple brands of stamp pad inks,followed by preprocessing using the Savitzky-Golay(SG)method.Subse-quently,a stamp pad ink type identification model was constructed based on the random forest(RF)algorithm.To validate the model's performance,the RF model was systematically compared with tra-ditional methods such as the backpropagation neural network(BP)and the multilayer perceptron(MLP).Additionally,grid search combined with five-fold cross-validation was utilized to optimize model parameters,thereby enhancing the model's performance and generalization capability.Exper-imental results demonstrated that the RF classification model(with 20 decision trees and 3 leaf nodes)outperformed the BP and MLP models across evaluation metrics including precision,sensitivity,specificity,and F1-score,achieving classification accuracies of 99.77％and 98.66％on the train-ing and test sets,respectively.The proposed method,integrating hyperspectral imaging technology with the RF algorithm,enables rapid,non-destructive,and accurate identification of stamp pad ink types,offering a novel technical reference for the analysis of atomic seal ink traces.关键词
高光谱成像/随机森林/原子印油/种类识别Key words
hyperspectral imaging technology/random forest/atomic oil/type distinction分类
化学引用本文复制引用
鲁晓权,陈航,马琨,李帆,张建强,张馨予,吴加权,李自超,张津豪,任慧慧..基于高光谱成像技术和随机森林的原子印油种类识别[J].分析测试学报,2025,44(11):2339-2345,7.基金项目
云南省科技厅基础研究专项(202101AU070011) （202101AU070011）
