Evaluation of Contrast Enhancement Methods for Face Recognition using SVM Classification
Santosh Kumar Jha, Prashant Kumar Jain, Prabhat Patel · 2025
Contrast enhancement techniques play a crucial role in improving the performance of face recognition systems. This paper evaluates two contrast enhancement methods, entropy-efficient adaptive DC coefficient scaling (DCCS) and Ant Colony Optimization (ACO)-based contrast enhancement, for face recognition using Support Vector Machine (SVM) classification. The proposed system involves loading a face image database, applying contrast enhancement to a query image, training an SVM classifier, extracting features from the enhanced query image, and conducting a performance evaluation. The DCCS method converts the image from RGB to LAB color space, applies a Discrete Cosine Transform (DCT), and adaptively maps the DC coefficients using a log-twicing function. The processed DCT blocks were merged and the images were converted back to RGB. An entropy-based decision-making process determines whether wavelet-based fusion should be applied to the original and DCCS-enhanced images. The ACO-based me the term emphasizes the luminance component, sets Simulated Annealing options, computes a relative range, and estimates a fineness function for ACO using correlation parameters derived from the mean, entropy, and filtered form of the image. The effectiveness of these contrast enhancement techniques was evaluated across diverse datasets and real-world scenarios, providing insights into their potential applications in biometric systems. The integration of sophisticated contrast enhancement algorithms with SVM based face image as binary classification has been trained. The DCCS based enhancement works superior then ACO.