Multiple Description Watermarking Based on Quantization Index Modulus Modulation

• Published in: Journal of Information Science and Engineering Vol. 23; no. 6; pp. 1785 - 1800
• Main Authors: 戴敏倫(Miin-Luen Day), 李素瑛(Suh-Yin Lee), 周義昌(I-Chang Jou)
• Format: Journal Article
• Language: English
• Published: Taipei 社團法人中華民國計算語言學學會 01.11.2007; Institute of Information Science, Academia sinica
• Subjects: Applied sciences; Artificial intelligence; Computer science; control theory; systems; Computer systems and distributed systems. User interface; Exact sciences and technology; Memory and file management (including protection and security); Memory organisation. Data processing; Pattern recognition. Digital image processing. Computational geometry; Software; watermarking; MDW; error-prone; QIMM; multiple description; Packet switching; Gaussian filtering; Image processing; Competitiveness; Median; Stéganography; Image compression; Signal quantization; Multiplicity; Signal sources; Safety; Transmission loss; Discrete cosine transforms; Transmission error; Digital watermarking
• ISSN: 1016-2364
• DOI: 10.6688/JISE.2007.23.6.9

In this paper, we study the problem of watermarking for error-prone transmission over unreliable network. We try to integrate an oblivious quantization index modulus modulation (QIMM) watermarking technique into the multiple description coding (MDC) framework and we call it multiple description watermarking technique (MDW). It is known that the balanced MDC encodes a signal source into multiple bitstreams (descriptions) of equal importance and equal data rate. Consider a traditional two-description case in a packet transmission network. The computation for watermark embedding is performed using either description. Once chosen, the corresponding values to be modulated for the other description are assigned with the same values as the just watermarked description. In the detection process, the embedded watermark could be extracted no matter either one or both descriptions are received. That is to say, the watermark is still detectable from MDW even with 50% packet loss. Furthermore, in the case of 50% packet loss, the resulting watermark from MDW is still robust to a variety of image processing attacks, including DCT based compression (JPEG), DWT based compression (JPEG-2000), Gaussian filtering, sharpening and median filtering. The experimental results confirmed the competitive performance and the effectiveness of the proposed scheme.