Exploiting Self-Embedding Fragile Watermarking Method for Image Tamper Detection and Recovery

Abstract

The increasingly rapid use of digital images makes data security technology become an important issue to ensure the integrity and ownership of the received image. Watermarking technique is a technique that ensures data security including for checking the authenticity of an image. Current developments, not only can test the authenticity of an image, but also can determine the location of damage while fixing it. The method proposed in this paper is through using the block-based technique of self-embedding fragile watermarking, where the watermark is selected from feature images representing the entire image. Four authentication bits and eight recovery bits are generated in each 2x2 non-overlapping block. Authentication bits are embedded in the three least significant bits (LSBs) of the block itself while recovery bits are embedded in the three LSBs of the corresponding mapped block. Our method is efficient as it only uses simple parity check operations and comparison between average intensities as image tamper detection and use the five most significant bits (MSBs) of 2x2 pixels in average for recovery. The experimental results of the tests on cropped images show that the proposed methods allow image recovery with acceptable visual quality better than some of state-of-the-art schemes.

Blind Robust and Self-Embedding Fragile Image Watermarking for Image Authentication and Copyright Protection with Recovery Capability

“…For each block, we adopt method [26] to generate two parts that will be respectively used for tampered area localization and content recovery. Assume that the number of MSB layers used for the generation of recovery bits is denoted as .…”

Video-based Data Hiding by Exploring Prediction Error and Histogram Shifting

“…Many RDH algorithms have been developed in various methods and applications, which are analyzed from different aspects, like in encrypted images [5], digital audio-based classification [6], trends of digital image-based data hiding [7], attack to cipher images [8], two-layer embedding [9], and predictor-based RDH [10]. In detail, the introduced methods include difference expansion (DE), histogram shifting (HS), and pixel value ordering (PVO), for various purposes, like file tampering detection [11] and watermarking [12]. Furthermore, RDH can be developed in various media, like image [7], audio [6], and video [13], with either spatial or frequency domain.…”

Fragile Image Watermarking Based on Bidiagonal SVD-LSB for Tamper Detection and Localization

“…There are also several complex attacks, these attacks are generally designed to hide interference, some of these attacks include vector quantization attack (VQA), collage attack (CA), content only attack (COA), constant feature attack (CFA) [2,10]. The ability of the fragile watermarking method to detect tampering can be determined by measuring tools such as the true positive rate (𝑇𝑃𝑅) or often also called the tamper detection rate (𝑇𝐷𝑅), true negative rate (𝑇𝑁𝑅) or false positive rate (𝐹𝑃𝑅), and precision (𝑝), where when these three measuring instruments have a value close to 1, it means that the quality of tamper detection is getting better [2,11,12]. The imperceptibility aspect also determines the quality of fragile image watermarking.…”

“…The most widely used instruments to measure imperceptibility quality are PSNR and SSIM [13]. Fragile image watermarking can be designed in the spatial domain as in research [11,[14][15][16] or transformation domain [1,17,18] or these combination [2,19,20]. Determination of methods and domains can influence the results and advantages of the method in certain aspects.…”