1. INTRODUCTION
In the information era, multimedia content is increasingly being produced and distributed in digital form. Meanwhile, with the wide availability of editing software, content modification is no longer a difficult job. A negative consequence is that the trustworthiness of multimedia data is often questioned. Therefore, technical solutions are desired to ensure the integrity of multimedia content. One possible solution is to use cryptographic hash functions. However, they are sensitive to any bit change, thus are not suitable for the multimedia domain where the same content often exists in many digital forms - different formats, different quality, etc., and they are still considered authentic even after moderate processing, such as common enhancement and slight geometric distortion. Apparently, it is impractical to compute and store conventional hash values for all possible cases. Therefore, it is expected to have hash algorithms that only depend on the content and tolerate content-preserving distortion, i.e., for the same or similar content, the hash is always the same or similar, regardless of the digital form; and the hash changes drastically only after the content is significantly changed. This motivates the research in perceptual hash algorithms. Basic components of a perceptual hash algorithm.