Attack and Improvement of the Fidelity Preserved Fragile Watermarking of Digital Images

Abstract

Recent fragile watermarking is applied to protect the content integrity, to detect and locate the tampered area, and to recover the tampered data of digital images. Since the watermark which is made up of a considerable amount of digital data distorts the protected image seriously, attention has been drawn to fidelity preservation of the watermarked image. A typical watermarking scheme with fidelity preservation was proposed by Lin et al., in which the superiority is that high fidelity of the watermarked image can be guaranteed while retaining the tamper-proof functionality. However, no recovery capability is provided in the original scheme; therefore it cannot meet the requirement of recent watermarking applications. In addition, the original scheme is insecure. Any part of the protected image can be replaced with an arbitrary image by some simple operations, and the illegal modification could not be detected. In this paper, we attack the original scheme first and then overcome the insecurity problem by using a modified weighted-sum function; we also design an improved scheme with recovery capability by using the vector quantization technology, while maintaining the superiority of the original scheme, i.e., fidelity preservation. Experiments verify the feasibility of the proposed method.