Scholarlink

The ability to identify individuals through fingerprints is a fundamental aspect of forensic science. However, with the emergence of advanced identity fraud techniques, traditional fingerprint identification systems face significant challenges. In this paper, Convolutional Neural Networks (CNNs) are utilized to classify altered fingerprints from real ones. Various deep learning architectures, including VGG19, VGG16, ResNet50, and InceptionV3, have been employed for classification. Additionally, a new CNN architecture is proposed, demonstrating superior performance in terms of accuracy and resource efficiency compared to existing models. This paper also explores the reconstruction of fingerprints from altered versions, enhancing the speed and accuracy of forensic investigation.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Altered Fingerprints Detection and Reconstruction for Forensic Investigation

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Srabani Biswas,
  • Sayani Chandra,
  • Suparna Biswas,
  • Ipsita Saha,
  • Srijit Bhowmick,
  • Tathagata Bhattacharjee,
  • Tanisha Bose

摘要

The ability to identify individuals through fingerprints is a fundamental aspect of forensic science. However, with the emergence of advanced identity fraud techniques, traditional fingerprint identification systems face significant challenges. In this paper, Convolutional Neural Networks (CNNs) are utilized to classify altered fingerprints from real ones. Various deep learning architectures, including VGG19, VGG16, ResNet50, and InceptionV3, have been employed for classification. Additionally, a new CNN architecture is proposed, demonstrating superior performance in terms of accuracy and resource efficiency compared to existing models. This paper also explores the reconstruction of fingerprints from altered versions, enhancing the speed and accuracy of forensic investigation.