On Improving Perceptual Image Hashing Using Reference Image Construction

Abstract

AbstractThis paper proposes an improved framework for perceptual image hashing, which is able to improve existing image hashing schemes. In our framework, hash sequence contains two parts. The first one is generated by an existing perceptual image hashing scheme on an input image. Then, a reference image is constructed by executing Wiener filtering on the input image. Subsequently, the same hashing scheme is executed on the reference image to obtain the second part. Finally, the two parts are combined together to form the improved hash. With the improved hash, the properties of main content of image can be fully extracted. Experimental results show that the performances of existing perceptual image hashing schemes, e.g., perceptual robustness, discrimination capability, can be improved using our framework.KeywordsDigital imagePerceptual hashingImprovement