Robust Image Fingerprinting via Distortion-Resistant Sparse Coding

Content fingerprinting recently emerges as an effective nonintrusive solution for copyright protection. Fingerprinting algorithm maps the perceptual contents of media file to an invariant digest, so that unauthorized copies can be identified via fingerprint comparison. This letter presents a distortion-resistant sparse coding strategy for image fingerprinting that simulates the hierarchical information processing flow of visual system. Sparse coding, which seeks a small set of atoms that can best represent input signal, helps fingerprinting algorithm detect the intrinsic visual features of image. However, the high freedom of atom selection makes sparse coding sensitive to distortion. In this letter, several measures are applied on sparse coding and dictionary learning to jointly ensure the invariance of fingerprint, such as imposing the neighborhood-priority principle on atom selection, regulating the layout of atoms, and forcing sparse codes to preserve the distance in the image space. Content identification performance of the proposed work was tested on a database of 219 000 images. The error rate of the proposed algorithm is at least ten times lower than state-of-the-arts, and satisfactory performance was observed even under extremely low bit budget.

I. Introduction

The emergence of social network makes content sharing unprecedentedly easy, while the problem of copyright infringement is also aggravated. A traditional technique for tackling this problem is digital watermarking. However, digital watermarking protects copyright at the price of degrading the perceptual quality of cover media. More importantly, it is not able to resolve the copyright dispute on the content where no watermark was embedded in advance. The robustness against distortion is another bottleneck in real applications. A media file may experience a bunch of distortions and transformations during its life-cycle, which significantly increases the difficulty of watermark detection. Content fingerprinting is a successful alternative to digital watermarking. Content fingerprinting summarizes the intrinsic perceptual characteristics of digital media into a short and invariant ID, such as human fingerprint, and the perceptual equality between media files can be determined by comparing their fingerprints. As a result, content fingerprinting has been widely adopted by social networks to filter the uploads that may cause copyright violation. Being a low-cost solution for content identification, content fingerprinting has also found applications in indexing, broadcast monitoring, recommendation, etc.