作者: Shijun Xiang , Jiwu Huang , Rui Yang
DOI: 10.1007/978-3-540-74124-4_7
关键词: Computer vision 、 Scale invariance 、 Artificial intelligence 、 Watermark 、 Time domain 、 Mathematics 、 Robustness (computer science) 、 Histogram 、 Digital watermarking 、 Audio signal 、 Audio watermark
摘要: In audio watermarking, the robustness to desynchronization attacks such as TSM (Time-Scale Modification) operations, is still an open issue. this paper, both mathematical proof and experimental testing show that histogram shape (represented relative relation in number of samples among three different bins) mean are two robust features attacks. Accordingly, a multi-bit watermarking algorithm based on statistical proposed by modifying histogram. The with equal-sized bins extracted from selected amplitude range referred mean, then relations groups neighboring designed carry watermark reassigning bins. watermarked signal perceptibly similar original one. Simulation results demonstrated hidden message very TSM, cropping, variety other distortions Stirmark Benchmark for Audio.