Theoretical Analysis of Information Watermarking in Wavelet-Based Video Compression

摘要

Embedding audio bits into images for transmission of video data alleviates the synchronization problem common in video transmission techniques. We continue work combining audio or other information bits and images into one file using digital watermarking techniques to correct the synchronization problem. The system compresses the file by using wavelet image coefficients and implementing bit plane coding. Our research encompasses incorporating five free variables into the watermark/compression technique. These variables are watermark robustness, number of coding iterations, number of image coefficients, number of watermarked information bits, and number of watermarked error correcting bits. By altering these variables, four measurements of the output change. The measurements are the information bit error rate, the image quality, the bit rate, and the amount of watermarked data. We theoretically demonstrate how the variables impact these measurements. Experimental results on real video data support our theoretical findings. By analyzing each video frame, an automated system is able to choose optimal values of the five variables to meet 5 specified measurement constraints.