Abstract

This paper presents a two-stage watermarking technique for image authentication adapting the advantages of vector quantization. In the proposed algorithm, robust watermark and semifragile watermark are embedded independently in two successive stages. Robust watermark and VQ enhance the security of the system by providing double protection to the designed system, while semifragile watermark helps in authenticating the received image. Watermarks of varying sizes are embedded in the cover image, and their performance is measured in terms of peak signal-to-noise ratio, weighted PSNR, and bit error rate. A threshold-based approach is suggested for identification of attacks as acceptable or malicious. The experimental results demonstrate the capabilities of the method in classifying attacks and correctly locating the tampered area. It is possible to detect and determine tampering with very high sensitivity. The present scheme outperforms previous algorithms in terms of imperceptibility, attack classification criteria, robustness feature, and tamper detection feature.