Spectral-hashing-based image retrieval and copy-move forgery detection

Abstract

Fast and effective retrieval of relevant images has become an important requirement in applications that make use of digital images. The numerous sophisticated yet simple to use image editing tools available have made image tampering an easy task. This paper focuses on using spectral hashing to address aforesaid issues. Sparsification of graph Laplacian using spectral hashing is used for effective image retrieval. Copy-move forgery detection is done using Spectral-hashing-based Polar Cosine Transform (PCT). The proposed algorithm also addresses the identification of tampered regions that have been subjected to additional post-tampering manipulations. Simulation results show that the spectral-hashing-based identification of similar patches gives better results compared with some of the existing hashing techniques.

Keywords:

• spectral hashing
• Polar Cosine Transform
• copy-move forgery detection