Robust image hashing for content authentication with tampering localization and image recovery

Abstract

Perceptual image hashing is a gifted key of image content authentication. Robustness and discrimination capability are two of the most important goal in image hashing algorithm. Designs an efficient image hashing incorporate with ring partition and invariant vector distance for enhancing rotational robustness and discrimination capability. The main intention of ring partition in image hashing is to construct unrelated to image rotation. The statistical features that are extracted from image rings in perceptual uniform color space, i.e., CIE L∗a∗b∗ color space, are rotation invariant and stable space. In specific, the Euclidean distance between perceptual statistic features vector is invariant to routinely used digital operations to digital images. (e.g., Brightness/contrast adjustment, JPEG compression, Salt and pepper noise), Which helps in build efficient image hash compact and discriminative. Execute experiments to evaluate the efficiency of image hashing algorithm is robust at digital operation to images. A wide range of geometric distortions and perceptual sensitivity to exposure malicious tampering. The tampering is localized and contents are recovered by new inpainting algorithm. It combines the strength of local and non local sparse representation in plenty of images under a systematic framework known Bayesian model averaging.