Imperceptible–visible watermarking for copyright protection of digital videos based on temporal codes

Abstract

Digital watermarking has emerged as an important tool for copyright protection of digital videos. In this paper, we introduce a robust and imperceptible video watermarking method performed in the base-band domain that is not computationally expensive. Two nonoverlapped watermarks are embedded in the host video frames. The first watermark is a grayscale image that represents the ownership of the video. First, this watermark is decomposed into a sequence of binary images called temporal codes by using a spatiotemporal masking model. Then, the temporal codes are embedded in the spatial domain of video frames by using the imperceptible–visible watermarking paradigm. Since it is well known that such a paradigm is computationally expensive, we employ a technique that considers the video scene as the basic processing unit to considerably increase the execution speed. The second watermark works as a supporting media to share data between embedding and exhibition stages and is embedded in the discrete cosine transform (DCT) with a very robust method based on quantization index modulation (QIM) under dither modulation and some spatiotemporal criteria of the human visual system (HVS). Computer simulations were conducted regarding robustness, imperceptibility, and time consumption to determine the feasibility of our proposal. Experimental results confirm that the combination of temporal codes and the imperceptible–visible watermarking paradigm is an innovation in the field that brings advantages such as simplicity, low computational complexity, and improved robustness and imperceptibility.