For many years information hiding techniques are playing major roles in achieving robustness against various malicious attacks on digital multimedia data of different application areas. The induction of digital watermarking has given these practices a new way in the field of information hiding to prevent the copyrighted multimedia data against various attacks. In this paper, we propose a generalised watermarking system which efficiently identifies various attacks in spatial, temporal, and spatio-temporal domain, viz. frame deletion, frame swapping, and frame copying, etc. Based on the timing information of each frame of a video, in the proposed scheme we first generate the unique watermarks (representing the timing information) in real time and then embed the auto generated watermark into 8 x 8 discrete cosine transform blocks of corresponding frames of a video. While extracting the watermarks from each frame of the watermarked video, any alterations in the sequence of the retrieved watermarks are the indication of the temporal attacks (frame deletion, etc.) in the copyrighted videos. Experiments have been conducted on self captured videos to analyze the performance of the proposed watermarking model. Performance of the proposed model has been measured in terms of the recall rate, i.e. ability of the proposed watermarking model to correctly detect the attacks in videos, quality of the watermarked videos and extracted watermarks using the quality metrics, Peak Signal to Noise Ratio (PSNR). Experimental results show the 100% recall rate in detection of the temporal attacks, if videos are manipulated in temporal domain only. Achieved recall rate is in between the range 86% and 91%, if videos are manipulated in spatio-temporal domain and it is 93%, if videos are spatially manipulated.
展开▼
机译：多年来，信息隐藏技术在实现针对不同应用领域的数字多媒体数据的各种恶意攻击的鲁棒性方面发挥着重要作用。数字水印的引入为这些实践提供了一种信息隐藏领域的新方法，以防止受版权保护的多媒体数据遭受各种攻击。在本文中，我们提出了一种通用的水印系统，可以有效地识别时空域，时空域中的各种攻击。帧删除，帧交换和帧复制等。基于视频的每个帧的时序信息，在提出的方案中，我们首先实时生成唯一的水印（代表时序信息），然后嵌入自动生成的水印分成对应视频帧的8 x 8离散余弦变换块。在从带水印的视频的每个帧中提取水印时，检索到的水印的顺序的任何更改都是对受版权保护的视频中临时攻击（帧删除等）的指示。已经对自拍视频进行了实验，以分析所提出的水印模型的性能。已根据召回率（即，所提出的水印模型使用质量度量，峰值信噪比（PSNR）正确检测视频中的攻击，水印视频的质量和提取的水印的能力）来衡量所提出模型的性能。 ）。实验结果表明，如果仅在时域中操作视频，则在检测时态攻击时100％的查全率。如果在时空范围内操作视频，则召回率在86％到91％之间;如果在空间上操作视频，则召回率在93％之间。
展开▼
