ABSTRACT

Fingerprinting scheme plays an important role for file sharing. In this paper, a novel method is introduced for fingerprinting an image. The proposed method uses wavelet and Principal Component Analysis (PCA) techniques for fingerprint generation. Fingerprinting is done two phases. During, first phase wavelet techniques are used to obtain low frequency representation of the test image. In second phase, PCA finds the features of the representation; set of fingerprint matrices can be created based on a proposed algorithm. Finally, each matrix combined with the low frequency representation to become a unique fingerprinted matrix. Size of the fingerprinted image is very small but it has most important information. Without this information, quality of the reconstructed image is very poor. We use different wavelet for fingerprint generation. Based on the performance, it is found that the DB6 is the best choice for fingerprint generation. Fingerprinted file is more suitable for distribution of file in peer-2-peer (p2p) networks.