Special Section on Security, Steganography, and Watermarking

Abstract

Motivated by the annual conference on Security, Steganography, and Water-marking of Multimedia Contents at the IS&T/SPIE Symposium on Electronic Imaging, the objective of this special section on Security, Steganography, and Watermarking is to present key is-sues in the areas of data authentication and data manipulation recognition, user authentication, detection of hid-den communication channels as well as trade-offs for robust digital water-marking and watermarking benchmark-ing approaches. We expect the special section to motivate these research fields and show the most relevant prob-lems and actual solutions. The special section consists of six reviewed papers. The papers address theoretical and practical issues with a high degree of innovation as well as in-teresting improvements of existing art or new ideas and research directions. The first paper, " Image manipulation detection " by Bayram, Avciba, Sankur, and Memon, discusses emerging tech-nologies in the digital data authentica-tion field by presenting techniques and methodologies for validating the au-thenticity of digital images and testing for the presence of doctoring and ma-nipulation operations. Three categories of forensic features are reviewed and the design of classifiers between doc-tored and original images is discussed. The three categories of features are the binary similarity measures between the bit planes, the image quality met-rics applied to denoised image residu-als, and the statistical features ob-tained from the wavelet decomposition of an image. The performance of clas-sifiers with respect to selected con-trolled manipulations as well as to un-controlled manipulations is analyzed. The tools for image manipulation de-tection are treated under feature fusion and decision fusion scenarios. The three forensic features were tested against the background of single ma-nipulations and multiple manipulations, as would actually occur in doctoring im-ages. The set of single-manipulation experiments shows that it is best to se-lect features from the general pool of all categories ͑feature fusion͒. In the second set of experiments with multiple manipulations, the best strategy was to use different types of classifiers ͑ex-perts͒, one per manipulation, and then fuse their decisions. Besides data authentication chal-lenges, user authentication applica-tions, especially biometric user authen-tication applications, are stepping into a new dimension with respect to the size and complexity, due to the deci-sion of many countries to introduce bio-metric travel documents. Along with this development, many problems have become increasingly relevant. The sec-ond paper, " Handwriting biometrics: is-sues of integration in identification documents and sensor interoperability " by Vielhauer, addresses two of these problems: security and privacy-preserving storage of biometric refer-ences and issues of sensor interoper-ability. For the first problem, security aspects of storage methods based on physical components, digital image processing, and centralized data stor-age techniques are discussed and re-viewed. Key research problems with the biggest impact on specified defi-ciencies in the field of secure physical storage in embedded devices are sum-marized. For the second problem, an intersensor evaluation methodology using the example of handwriting is in-troduced and extended by a new verifi-cation algorithm called BioHash. The concept is experimentally cross-validated and main conclusions are drawn. For example, the degradation of recognition accuracy in online signa-ture verification may reach up to a fac-tor of 4.4 for nonskilled forgeries, when cross-sensor verification is involved. Al-ternative semantic classes such as in-dividual symbols show a similar ten-dency, although the absolute degrad-ation factor appears to be lower. Se-mantic alternatives to signatures such as pass phrases or hand-drawn sym-bols have been shown to be well suited for online handwriting biometrics. The third paper, " Performance study of common image steganography and steganalysis techniques " by Kharrazi, Sencar, and Memon, investigates the performance of state-of-the-art univer-sal steganalyzers proposed in the lit-erature. The paper presents