Real-time perceptual watermarking architectures for video broadcasting

  • Authors:
  • Saraju P. Mohanty;Elias Kougianos

  • Affiliations:
  • NanoSystem Design Laboratory (NSDL),1 University of North Texas, Denton, TX 76207, USA;NanoSystem Design Laboratory (NSDL),1 University of North Texas, Denton, TX 76207, USA

  • Venue:
  • Journal of Systems and Software
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

Existing secure embedded systems are primarily cryptography based. However, for effective Digital Rights Management (DRM) of multimedia in the framework of embedded systems, both watermarking and cryptography are necessary. In this paper, a watermarking algorithm and corresponding VLSI architectures are presented that will insert a broadcaster's logo into video streams in real-time to facilitate copyrighted video broadcasting and Internet protocol television (IP-TV). The VLSI architecture is prototyped using a hardware description language (HDL) and when realized in silicon can be deployed in any multimedia producing consumer electronics equipment to enable real-time DRM right at the source. The watermark is inserted into the video stream before MPEG-4 compression, resulting in simplified hardware requirements and superior video quality. The watermarking processing is performed in the frequency (DCT) domain. The system is initially simulated and validated in MATLAB/Simulink^(R) and subsequently prototyped on an Altera^(R) Cyclone-II FPGA using VHDL. Its maximum throughput is 43frames/s at a clock speed of 100MHz which makes it suitable for emerging real-time digital video broadcasting applications such as IP-TV. The watermarked video is of high quality, with an average Peak-Signal-to-Noise Ratio (PSNR) of 21.8dB and an average Root-Mean-Square Error (RMSE) of 20.6.