Biometric Image Integrity Authentication Using SVD and Fragile Watermarking

This MATLAB code implements the methodology presented in the research paper "Biometric Image Integrity Authentication Based on SVD and Fragile Watermarking." The research focuses on developing an integrity verification system for biometric images using Singular Value Decomposition (SVD) and fragile watermarking techniques. The implementation follows a structured approach where the input image first undergoes SVD decomposition to extract essential feature vectors. The code includes functions for matrix decomposition using MATLAB's built-in svd() function, which separates the image into orthogonal matrices and singular values. These extracted features are then compared with original image characteristics to verify integrity through calculated similarity metrics. Concurrently, the implementation incorporates fragile watermarking algorithms that embed authentication data directly into the image pixels. The watermark embedding process utilizes bit-plane manipulation techniques to insert verification information without significantly affecting visual quality. Corresponding extraction functions can detect even minor tampering by analyzing the embedded watermark integrity. Experimental validation within the code demonstrates robust performance against common attacks including noise addition (Gaussian, salt-and-pepper) and image compression artifacts. The authentication system shows high sensitivity to unauthorized modifications while maintaining low false-positive rates. This comprehensive MATLAB implementation serves as both an educational resource for understanding SVD-based authentication mechanisms and a practical tool for researchers developing advanced biometric security systems. The code includes detailed comments explaining key algorithmic steps, parameter configurations, and performance evaluation metrics to facilitate easy adaptation and extension for related research applications.