Application of visible-near infrared hyperspectral imaging for detecting latent washed bloodstains on Colored fabrics
Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy · 2026-11 · 361 · 128093
DOI: 10.1016/j.saa.2026.128093 ↗Kurz gefasst
Fachartikel in der Zeitschrift Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2026-11. Verfasst von Ligen Ai und 3 weiteren Autor:innen. Beteiligte Einrichtung: Affiliated Hospital of Inner Mongolia Minzu University, Inner Mongolia Minzu University, No.1742 East Huolinhe Street, Tongliao, 028000, Inner Mongol…. Hauptthemen laut MEDLINE: Blood Stains, Hyperspectral Imaging, Textiles. Die Arbeit nennt 32 Literaturangaben.
Abstract (Originalsprache)
Clothing stained with blood is a key form of evidence in forensic investigations, but washing and the presence of colored fabric backgrounds significantly increase the difficulty of bloodstains detection. Visible-near infrared hyperspectral imaging (VNIR-HSI), by integrating imaging and spectral capabilities, offers a promising solution. In this study, bloodstain samples were prepared on red, blue, purple, and white cotton fabrics, washed with three common detergents, and then detected using a VNIR-HSI system. To improve detection of latent washed bloodstains, pseudocolor visualization, spectral background correction and machinelearning were applied to the VNIR-HSI data. The results showed that washed bloodstains on colored fabrics were difficult to recognize with the naked eye because of strong substrate masking, whereas pseudocolor visualization markedly improved stain contrast and localization. Spectral analysis indicated that fabric background color had a much greater effect on spectral response than inter-individual variation or detergent type. After spectral background correction (standard normal variate-normalization), blood-related features were effectively enhanced, and the classification models achieved high performance, with SVM-RBF showing the best overall results (achieved an accuracy of 0.96). These findings demonstrate that VNIR-HSI, combined with pseudocolor visualization, spectral background correction and machine learning, can reliably detect latent washed bloodstains on colored fabrics and provides a promising rapid, visual and non-destructive method for forensic screening of washed colored clothing.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Affiliated Hospital of Inner Mongolia Minzu University, Inner Mongolia Minzu University, No.1742 East Huolinhe Street, Tongliao, 028000, Inner Mongolia Autonomous Region, China.
- Department of Pediatrics, School of Clinical Medicine, Inner Mongolia Minzu University, No.536 West Huolinhe Street, Tongliao, 028000, Inner Mongolia Autonomous Region, China.
- Department of Forensic Medicine, School of Basic Medical Sciences, Central South University, No.172 Tongzipo Road, Changsha, 410013, Hunan Province, China.
- Department of Forensic Medicine, School of Basic Medical Sciences, Central South University, No.172 Tongzipo Road, Changsha, 410013, Hunan Province, China. Electronic address: 15802644780@163.com.
Zitieren
APA
Ai, L., Wang, H., Ma, Y., & Wen, D. (2026). Application of visible-near infrared hyperspectral imaging for detecting latent washed bloodstains on Colored fabrics. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 361, 128093. https://doi.org/10.1016/j.saa.2026.128093
Vancouver
Ai L, Wang H, Ma Y, Wen D. Application of visible-near infrared hyperspectral imaging for detecting latent washed bloodstains on Colored fabrics. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy. 2026;361:128093. doi:10.1016/j.saa.2026.128093
BibTeX
@article{Ai2026,
author = {Ai, Ligen and Wang, Hongzhe and Ma, Yajie and Wen, Dan},
title = {Application of visible-near infrared hyperspectral imaging for detecting latent washed bloodstains on Colored fabrics},
journal = {Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy},
year = {2026},
volume = {361},
pages = {128093},
doi = {10.1016/j.saa.2026.128093}
}
EUVIMED