Optimized face recognition algorithm using radial basis function neural networks and its practical applications.

Area of Science:

• Computer Vision
• Biometrics
• Artificial Intelligence

Background:

• Face recognition systems require robust preprocessing and accurate identification methods.
• Existing methods often face challenges with image quality variations and dimensionality.
• Hybrid approaches integrating feature extraction and advanced classification are needed.

Purpose of the Study:

• To propose a novel hybrid method for enhanced face recognition.
• To improve facial image preprocessing using Active Shape Model (ASM) and Principal Component Analysis (PCA).
• To develop an accurate face identification model using Polynomial-based Radial Basis Function Neural Networks (P-RBF NNs).

Main Methods:

• Image acquisition using CCD camera, enhanced by AdaBoost and Histogram Equalization (HE).
• Preprocessing combining ASM for contour extraction and PCA for dimensionality reduction.
• Recognition using improved RBF NNs with fuzzy rules, Fuzzy C-Means (FCM) clustering, and Differential Evolution (DE) for parameter optimization.

Main Results:

• The proposed hybrid method effectively extracts facial features and reduces image dimensionality.
• The P-RBF NNs architecture demonstrates capability in identifying unique facial patterns.
• Optimization using Differential Evolution (DE) enhances network performance and recognition rates.

Conclusions:

• The hybrid face recognition approach integrating ASM, PCA, and P-RBF NNs offers a promising solution.
• The method shows potential for high performance and accurate recognition rates in facial identification.
• Further validation on diverse datasets is recommended to confirm generalizability.