Scholarlink

<p>The biometric recognition plays secure banking identity verification, in border control for efficient and reliable access management, in smartphone authentication user access it with enhanced security in real world. This paper documents a biometric recognition of Multi-Modal Adaptive Biometric Recognition Network (MABRecNet) designed to improve airport security, accurate, and reliable. MABRecNet addresses core challenges in biometric traits, enhanced feature extraction, overfitting prevention, scalability and robustness across datasets. The&#xa0;methodology starts with preprocessing to enhance image quality and increase variability using data augmentation. The Adaptive PolyNet is core model in multi-path dynamic structure capable of capturing low and high-level features from biometric traits, adaptable for changes in lighting conditions, pose, and user behaviour. Its adaptive branch activation and attention mechanisms pay attention on key feature vectors that enhance the recognition accuracy. Besides, Late Score Fusion-based FractalNet Deep CNN mines various feature vectors with different modalities to be fused at late score for better recognition performance. In addition, the MSMO algorithm aims at optimizing feature selection and dimensionality reduction through a balance of relevance and redundancy. This provides computational efficiency for real-time applications. The optimized feature set is processed in the Feature-Wise deep convolution neural network FWDCNN module of MABRecNet, which captures unique characteristics of combined biometric traits. This brings out the correct recognition result, hence improving recognition accuracy and robustness in multi-modal biometric systems. MABRecNet shows high performance with 99% Accuracy, 98.5% Precision, 98.8% mAP, 98.5% Recognition rate and, 97.5% GAR recognition. The framework enhances the biometric identification system on security and efficiency, even at large scales deployments.</p>

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

PolyNet-FractalNet deep feature fusion framework with modified spider monkey optimization for multi-modal biometric recognition

  • 期刊论文

  • 发表日期：

    2025年6月10日

查看全文
  • Laxman Singh,
  • Ashish Kumar,
  • Richa Golash

摘要

The biometric recognition plays secure banking identity verification, in border control for efficient and reliable access management, in smartphone authentication user access it with enhanced security in real world. This paper documents a biometric recognition of Multi-Modal Adaptive Biometric Recognition Network (MABRecNet) designed to improve airport security, accurate, and reliable. MABRecNet addresses core challenges in biometric traits, enhanced feature extraction, overfitting prevention, scalability and robustness across datasets. The methodology starts with preprocessing to enhance image quality and increase variability using data augmentation. The Adaptive PolyNet is core model in multi-path dynamic structure capable of capturing low and high-level features from biometric traits, adaptable for changes in lighting conditions, pose, and user behaviour. Its adaptive branch activation and attention mechanisms pay attention on key feature vectors that enhance the recognition accuracy. Besides, Late Score Fusion-based FractalNet Deep CNN mines various feature vectors with different modalities to be fused at late score for better recognition performance. In addition, the MSMO algorithm aims at optimizing feature selection and dimensionality reduction through a balance of relevance and redundancy. This provides computational efficiency for real-time applications. The optimized feature set is processed in the Feature-Wise deep convolution neural network FWDCNN module of MABRecNet, which captures unique characteristics of combined biometric traits. This brings out the correct recognition result, hence improving recognition accuracy and robustness in multi-modal biometric systems. MABRecNet shows high performance with 99% Accuracy, 98.5% Precision, 98.8% mAP, 98.5% Recognition rate and, 97.5% GAR recognition. The framework enhances the biometric identification system on security and efficiency, even at large scales deployments.