Video watermarking method based on visual attention mechanism

Abstract

The invention discloses a video watermarking method based on a visual attention mechanism. Binary-system watermark information is embedded into a DCT domain in a visual non-saliency area of an MPEG-2 video key frame is selected to achieve the purpose of video copyright protection. The video watermarking method comprises the following steps of decoding a carrier video and obtaining the video key frame I; adopting a Koch and Itti visual remarkable model to analyze visual saliency of the video key frame and dividing a visual saliency area; selecting the visual non-saliency area human eyes are not too sensitive to, performing fixed-step-length odd-even quantization index modulation on a DCT direct current coefficient of luminance component and embedding watermark information; analyzing the visual saliency of the video key frame with embedded watermarks, judging a watermark embedding area and extracting the watermark information according to watermark embedding rules. The visual attention feature of the human eyes is effectively utilized in the video watermarking method, the invisibility of the video watermarks is improved, the video watermarking method has good robustness on common video attacks, and contradiction between the watermark invisibility and the robustness is effectively relieved.

Description

technical field [0001] The invention relates to a multimedia information copyright protection method, in particular to a video watermarking method based on a visual attention mechanism, and belongs to the technical field of information security. Background technique [0002] Digital watermarking is one of the means to effectively resolve intellectual property disputes. Video is the main form of user-generated content (UGC) on the Internet, and the convenient way of obtaining it has spawned many infringements and piracy. Therefore, how to use video watermarks to protect the legitimate rights and interests of copyright owners has become a hot research topic at present, weighing the invisibility of watermarks The contradiction between robustness and robustness is still the common goal of all watermarking algorithms. [0003] Video is a series of time-related image sequences, so many image watermarking algorithms are also applied to video watermarking [1] . At the same time, ...

AI Technical Summary

Problems solved by technology

At the same time, due to the particularity of the video itself: large amount of data, complex data structure, real-time requirements, compression coding standards, uneven motion areas and non-motion areas, and specific types of attacks, the development of video watermarking is relatively lagging behind. [2]

Different video coding standards have limitations on video watermarking. Videos corresponding to the mature MPEG-2, MPEG-4 and the latest H.264 / AVC video coding standards require different watermarking schemes, which are embedded in the video. Formidable Technical Barriers to Robust Invisible Watermarking

[0004] visual attention mechanism [3] There have been many studies on image watermarking, but due to the complexity of video, few algorithms use human vision models in video watermarking, and most existing algorithms only embed watermarks in the transform domain or in motion vectors, so The quality of the obtained video often changes significantly, and the human eye can easily perceive the distortion of the video, which cannot satisfy the invisibility of the watermark

Images