Palm Vein Identification based on hybrid features selection model (2007.16195v1)

Abstract: Palm vein identification (PVI) is a modern biometric security technique used for increasing security and authentication systems. The key characteristics of palm vein patterns include, its uniqueness to each individual, unforgettable, non-intrusive and cannot be taken by an unauthorized person. However, the extracted features from the palm vein pattern are huge with high redundancy. In this paper, we propose a combine model of two-Dimensional Discrete Wavelet Transform, Principal Component Analysis (PCA), and Particle Swarm Optimization (PSO) (2D-DWTPP) to enhance prediction of vein palm patterns. The 2D-DWT Extracts features from palm vein images, PCA reduces the redundancy in palm vein features. The system has been trained in selecting high reverent features based on the wrapper model. The PSO feeds wrapper model by an optimal subset of features. The proposed system uses four classifiers as an objective function to determine VPI which include Support Vector Machine (SVM), K Nearest Neighbor (KNN), Decision Tree (DT) and Na\"ive Bayes (NB). The empirical result shows the proposed system Iit satisfied best results with SVM. The proposed 2D-DWTPP model has been evaluated and the results shown remarkable efficiency in comparison with Alexnet and classifier without feature selection. Experimentally, our model has better accuracy reflected by (98.65) while Alexnet has (63.5) and applied classifier without feature selection has (78.79).

• The paper introduces the 2D-DWTPP model that combines 2D DWT, PCA, and PSO to optimize palm vein pattern extraction.
• It employs preprocessing techniques such as adaptive histogram equalization and image inversion to enhance vessel visibility.
• Empirical results on the PUT Vein Database demonstrate a 98.65% recognition accuracy with SVM, outperforming conventional CNN models.

Analysis of Palm Vein Identification Using a Hybrid Feature Selection Model

The paper presents an advanced approach to palm vein identification, a biometric technique known for its robustness due to the unique, non-intrusive, and hard-to-forge characteristics of palm vein patterns. This research introduces a hybrid model named 2D-DWTPP, combining two-dimensional Discrete Wavelet Transform, Principal Component Analysis (PCA), and Particle Swarm Optimization (PSO) to enhance palm vein pattern extraction and reduce feature redundancy, which are critical challenges in biometric recognition systems.

Methodological Framework

The proposed model is structured into key phases: preprocessing, feature extraction, feature reduction, and classification. In preprocessing, palm vein images undergo adaptive histogram equalization and conversion to negative images to improve vessel visibility. Feature extraction employs the 2D wavelet transform, which efficiently highlights patterns by decomposing images into high and low-frequency components.

Subsequently, PCA is utilized to minimize feature dimensionality and redundancies, a necessary step to optimize the feature selection process. PSO is employed to iteratively refine the feature subset, leveraging its inherently efficient search capabilities by mimicking social behaviors in particle groups. This selection mechanism lays the groundwork for employing classifiers such as SVM, KNN, Decision Tree, and Naïve Bayes, with SVM demonstrating superior performance.

Empirical Evaluation and Results

The model's performance is validated using the PUT Vein Database, comprising images of varied sessions from both left and right hands. The experimental outcomes indicate a substantial enhancement in recognition accuracy, notably achieving 98.65% accuracy with the left-hand data using SVM. In comparison, pre-trained CNN models like AlexNet achieve substantially lower accuracy, demonstrating the efficacy of the proposed feature selection framework in handling high-dimensional datasets.

Implications and Future Directions

The results underscore the capability of the 2D-DWTPP model to improve machine learning model performance by preprocessing data to enhance feature datasets. This research suggests a robust pathway forward for biometrics applications where a high precision of feature extraction and selection is critical. Particularly, the demonstrated success with SVM in this context suggests a strong alignment between linear feature reduction techniques and support vector models.

Future research could focus on refining feature extraction and selection algorithms to reduce computation times while maintaining or improving classification accuracy. Integrating more advanced AI techniques or hybrid models could further enhance system capabilities, potentially opening new applications in security-sensitive environments that demand high accuracy and discriminatory power.

In conclusion, this paper contributes significant advancements in palm vein identification technology through an innovative hybrid feature selection model, showcasing valuable improvements over conventional classification methods in biometric systems.