Automatic Privacy Protection Based on Haar Cascades and Spectral Domain Watermarking
Conference: nDS '13 - Proceedings of the 8th International Workshop on Multidimensional Systems
09/09/2013 - 09/11/2013 at Erlangen, Deutschland
Proceedings: nDS '13
Pages: 6Language: englishTyp: PDF
Personal VDE Members are entitled to a 10% discount on this title
Authors:
Sablik, Thomas; Velten, Joerg; Kummert, Anton (Faculty of Electrical, Information and Media Engineering, University of Wuppertal, 42097 Wuppertal, Germany)
Abstract:
A new system for automatic privacy protection in digital media, based on Haar cascades and spectral domain watermarking is described in the present paper. In a first step faces are detected by means of Haar cascades. Integral images are used to speed up calculations and the detection. Multiple detections of one face are combined. Succeeding steps comprise embedding the data into the image using spectral domain methods and deleting the area of privacy. An extended Discrete Haar Transformation is developed to transform arbitrary matrices. A Spread Spectrum method is used to embed the size and position of the faces into the cover image. Different methods for embedding regarding their fidality and performance are compared. Additionally the performance of the methods after cropping the images is presented.