Secure Oblivious Hiding, Authentication, Tamper Proofing, and Verification Techniques

摘要

In this report, we describe an algorithm for robust visual hash functions with applications to digital image watermarking for authentication and integrity verification of video data and still images. The robust image digest can also be used as a search index for efficient database searches. The hash function depends on a parameter K (a secret key) in a sensitive manner and on the image in a robust, continuous manner. The hash function always returns the same N bits from any image of arbitrary size. The bits obtained from two different images or for two different keys K will generally be different (uncorrelated). However, for the same key K, two images that can be matched after applying gray scale operations, such as lossy compression, recoloring, filtering, noise adding, gamma correction, and simple geometrical operations including rotation and scaling, the extracted N-tuple will be almost the same. We also explain how the extracted N-tuple can be further utilized for synthesizing a Gaussian sequence that gradually changes with increasing number of errors in the extracted bits. Thus, the robust hash function can be used for generating pseudo-random watermark sequences that depend sensitively on a secret key yet continuously on the image. This robustness enables us to construct watermarks that depend on the original unwatermarked image in a non-trivial manner while making it possible to recover the watermark without having to access any information about the original image (oblivious watermarking). Such watermarks play an important role for authenticating videos or still images taken with a digital camera.