A Counter-Geometric Distortions Data Hiding Scheme Using Double Channels in Color Images

• Published in: Digital Watermarking pp. 42 - 54
• Main Authors: Xue, Gang, Lu, Peizhong, Wang, Jinlian
• Format: Book Chapter
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg 2005
• Series: Lecture Notes in Computer Science
• ISBN: 3540248390; 9783540248392
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-540-31805-7_4

This paper presents a new approach for data hiding with robustness to global geometric distortions. The global geometric distortions can be described by a 6-parameters affine transformation. Our scheme is designed for color images, which is combined with error-correcting code, double-channels steganography, feature point extraction, and triangle warping. Two color spaces of RGB images are considered two independent channels, one for synchronization information, and the other for a large amount of hiding data. The synchronization information consists of the coordinates of triangles’ centers, cyclic redundancy check bits, and parity check bits of convolutional codes. Global geometric distortions can be estimated successfully by least square method and K-means method. Then a large amount of data with low bit-error rate can be decoded by SOVA algorithm of Turbo coding after the geometric adjustment. We also improve the method for feature point extraction. Simulation results show that our scheme is robust to rotation, scaling, translation, cropping, shearing, and so on.