Quality evaluation and enhancement of audiovisual signals based on digital watermarking

图像、视频、语音和音频的质量测量对于当今的音频和视频广播、电话网络、传输控制和电子商务至关重要，这主要是因为质量既是客户满意度的关键决定因素，也是传输条件的关键指标。虽然主观质量评价是最可靠的视听质量评估方法，但它非常不适合在线监测和广播应用，并且非常昂贵和耗时。现有的客观质量评价指标都需要原始信号来评价视听质量。原始信号不适用于某些重要应用。提出了一种基于数字水印的视听质量评价方法。提出的研究的基础是，仔细嵌入在视听信号中的水印将经历相同的失真作为主机视听信号做。通过参考真实检测率和包括人类主观评价的特定常用质量度量之间的理想映射曲线，可以基于所提取的水印的劣化来测量视听质量。基于水印的质量评估不需要原始信号，并且可以非常准确。为了在Internet或无线网络上真实的实时、高质量地传输音频和视频数据，开发简单、鲁棒的错误隐藏和编码策略非常重要。使用音频/视频水印来包括附加的感知上透明的信息以帮助恢复分组或帧丢失的想法也可以导致若干优点。因此，建议的研究的一个组成部分是调查使用基于水印的数据包丢失隐藏方案的视听信号，压缩恢复传输的视频，自适应编码和传输的语音信号在ad-hoc无线网络。

The quality measurement of image, video, speech and audio is of critical importance for today's audio and video broadcasting, telephony networks, transmission control, and e-commerce, mainly because quality is both a key determinant of customer satisfaction and a key indicator of transmission condition. Although subjective quality evaluation has been the most reliable method of audiovisual quality assessment, it is highly unsuitable for online monitoring and broadcasting applications, and is also very expensive and time consuming. The current existing objective quality evaluation metrics require original signal to evaluate audiovisual quality. Original signal is not available for some important applications. We propose to evaluate audiovisual quality based on digital watermarking. The basis of the proposed research is that the carefully embedded watermark in an audiovisual signal will undergo the same distortion as the host audiovisual signal does. The audiovisual quality can be measured based on the degradation of the extracted watermark, by referring to an ideal mapping curve between the true detection rate and a particular commonly used quality metric including human subjective evaluation. The watermarking based quality evaluation does not need the original signal and can be very accurate. To deliver audio and video data over the Internet or wireless networks in real time with high quality, it is important to develop simple, robust error-concealment and coding strategies. The idea of including additional perceptually transparent information using audio/video watermarking to help for the recovery of packet or frame losses can also lead to several advantages. A component of the proposed research is therefore to investigate the use of watermarking-based packet loss concealment schemes for audiovisual signals, compression restoration in transmission of video, and adaptive coding and transmission of speech signal over ad-hoc wireless networks.