Browsing by Author "Guide - Dr. P. Amudha"

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Finger Vein Biometric Authentication using Deep Learning Techniques with Hybrid Labelling and Data Augmentation
    (Avinashilingam, 2025-05) Amitha Mathew; Guide - Dr. P. Amudha
    Biometric systems, designed to identify individuals through unique biological or behavioural characteristics, have become critical in modern security and identity verification applications. Finger vein authentication, which utilizes the unique vascular patterns within a person’s finger, offers intrinsic security due to the internal and difficult-to-replicate nature of vein patterns. However, the accuracy and reliability of finger vein recognition systems are heavily influenced by the quality of feature extraction techniques and environmental factors, such as motion artifacts, lighting, and finger positioning. This study focuses on developing a motion-tolerant deep learning model for finger vein recognition by employing advanced image labelling and dataset augmentation techniques. Traditional finger vein recognition systems rely on handcrafted features and classical machine learning algorithms, which face limitations in adaptability and accuracy under variable conditions. Recent advancements in deep learning have demonstrated the potential to automatically extract complex patterns from raw image data, addressing these challenges. In the first phase, a modified UNET and VGG16 model are used for recognition and the performance have been compared on the benchmark datasets, SDUMLA-HMT and THUFV. The U-Net, originally developed for biomedical image segmentation, has been modified by incorporating a fully connected layer, extending its capabilities to include classification tasks. VGG16 outperformed UNET across multiple metrics, with its deep feature extraction capabilities enabling it to capture intricate vein patterns and achieve higher classification accuracy. VGG16 also demonstrated superior precision and recall, minimizing false negatives and improving robustness in practical applications. The second phase explores the impact of using a hybrid labelling algorithm on the VGG16 model’s performance. By accurately labelling finger vein images, the model is able to learn detailed vein patterns more effectively, leading to improved recognition accuracy and reduced over fitting. This phase emphasized the importance of creating diverse and representative datasets to enhance the model’s generalization ability. In the third phase, various dataset augmentation techniques are investigated, including conventional transformations (rotation, scaling, and flipping) and an advanced approach using Generative Adversarial Networks (GANs). The experimental results revealed that the augmentation techniques significantly enhanced the diversity and quality of the training dataset, further improving the VGG16 model’s recognition performance. In the final phase, an adaptive finger vein recognition system that integrates VGG16 for feature extraction with Long Short-Term Memory (LSTM) for sequence learning has been developed. This motion-tolerant model is designed to handle real-world conditions, including motion artifacts and environmental variations, ensuring accurate recognition even when the finger is not perfectly still. The architecture was tested on the THUFV and SDUMLA-HMT datasets, demonstrated improvements in accuracy and robustness. This research demonstrates the effectiveness of combining VGG16 and LSTM architectures along with labelled and augmented dataset, to develop a reliable and scalable motion-tolerant finger vein recognition system. The findings contribute to the advancement of biometric authentication technologies, offering a robust solution for real-world applications.
  • No Thumbnail Available
    Item
    Hybrid Transfer Learning Models for Video Anomaly Detection in Surveillance Systems
    (Avinashilingam, 2025-03) Sreedevi R Krishnan; Guide - Dr. P. Amudha
    In a modern civilized community, public safety is of prime importance, and the detection of anomalous events has become a vital factor for a successful security system. Conventional Video Surveillance (VS) methods are inadequate for identifying anomalous events by themselves. It is due to their inability to analysis large sequential data in dynamic environments. Video Anomaly Detection (VAD) has undergone swift development with the emerging Artificial Intelligence (AI) technologies. The research addresses the challenges of conventional VAD and proposes hybrid Deep learning (DL) models using transfer learning techniques for an efficient VAD system in dynamic environments. The application range of VAD is not limited to but includes social, commercial and industrial surveillance systems. Traffic management in urban areas, crowd management, emergency response and resource optimization are VAD's key application fields. The wide requirement for intelligent VAD inspires the design and development of reliable and efficient video surveillance systems capable of automating Anomaly Detection (AD) with minimal human intervention. The study develops and evaluates four hybrid models for VAD using Deep Learning techniques based on transfer learning to obtain improved performance. The first research phase proposed a CNN-YOLO hybrid model capable of anomaly detection. This model uses CNN for model training and modified YOLOv4 for object detection, ensuring accurate and high-speed anomaly detection. This model processes a single random frame out of 100 input frames and yields a faster response. The CNN-YOLO have high accuracy and faster response but being a small model, it samples a random frame input only. To overcome this limitation and the inability of sequential video processing of the CNN-YOLO model, a hybrid model comprised of Residual Network (ResNet) and Long Short-Term Memory (LSTM) was executed in the second phase. This model can execute feature extraction and sequential information processing in more than thousands of video frames. ResNet-50 is employed for spatial feature extraction and LSTM to capture temporal relationships of the input video data even though this hybrid model enhances detection capability but has low accuracy, efficiency and generalization skill due to overfitting. In the third research phase, a segmentation-based anomaly detection technique is implemented, reducing overfitting. This model is constituted by hybridizing Improved UNet (IUNet) with the Cascade Sliding Window Technique (CSWT). In the IUNet-CSWT hybrid model, standard convolutional layers of IUNet are replaced with a ConvLSTM for spatiotemporal feature extraction. CSWT estimates the anomaly score of the input video and thus classifies it to normal and anomalous events. This model is equipped for processing complex patterns since it has an effective equilibrium between generalization skill and precision. A low false positive rate and high detection accuracy make the model work effectively in crowded environments. The fourth phase implemented a Hierarchical Multiscale-CNN with LSTM model, enhancing multi-scale feature identification and temporal data analysis. This model can work efficiently even in low-resolution video utilizing a Bilateral-Wave Denoise Technique. Multi-scale CNN is augmented by a Spatial Pyramid Pooling (SPP), which enhances the feature extraction. The performance of this model outperforms all the other models.