| 摘要 |
A channel is inserted into a sequence of frames for an image, by varying one or more display characteristics of the resulting image across the display in accordance with a spatio-temporal pattern applied to, for example, successive lines of each modified frame. In some cases, a reverse of the pattern is applied to a successive frame, and the frames modified are preferably B-frames if the image may be encoded in accordance with an MPEG standard. The channel may be employed to watermark the image, and the watermark may correspond to the presence of the channel or may be data in accordance with watermark information carried within the channel. For most display devices, display variations are minimized during the design process, but human viewers may still tolerate and accept subtle variations in a displayed image. Slightly changing one or more display characteristics in accordance with watermark information allows for watermarking of the image since viewers may not be aware that display characteristics are changing. For example, the blue color component of the video signal may be modified in accordance with watermark information defined as W(x) and its complement [1-W(x)] in counter-phase rotations of successive lines in pairs of frames. The blue color component of the video signal may be modified in those regions of the image declared to be relatively still and with sufficient level of blue-lightness. If the video signal is not encoded, the watermark information may alter each pre-selected pixel value's least significant bit. If the video signal is encoded, such as into the frequency domain as specified in the MPEG standard, the corresponding (2,2) DCT coefficient in pairs of B-frames may be modified in accordance with watermark information. The watermark information is applied in relatively still portions of the image with sufficient blue-lightness level. A receiver includes a watermark detector that has a priori information about the watermark. The watermark detector examines specific regions of the displayed image over time, and tests the regions for the watermark during time intervals when the image remains relatively still. The receiver calculates the difference in pixel values in still regions between the received test frame and either a) corresponding pixel values in the non-modified image frame or b) the corresponding pixel values in the successive frame of a pair.
|
