LBP-based fragile watermarking scheme for image tamper detection and recovery

This paper proposes a fragile image watermarking scheme with recover ability based on local binary pattern (LBP). The local binary pattern operator is mostly used to extract localized spatial features for texture analysis, image classification, and pattern recognition. According to spatial relations of a natural image, a local binary pattern can be utilized to represent the localized relations of a pixel with its neighborhood. Each pixel measured by the LBP operator can be obtained its own local binary pattern as representation of local spatial relations. The proposed scheme utilizes the LBP operator to generate authentication data which are embedded into each image block with 3 × 3 pixels size for tamper detection and recovery. The recovery information is obtained by calculating the mean value of each image block, and then the mean value is converted into a binary string which is embedded into eight neighboring pixels' LSBs of each image block for image recovering. From the experimental results, the proposed scheme has efficient tamper detection, better image quality and recovering ability than Zhang and Shih's LBP based watermarking scheme.