在離散小波轉換域中結合時間擴展回聲與有理抖動調變以增進音訊盲浮水印效能

• Main Authors: Lai, Sheng-Yu, 賴聖瑜
• Other Authors: Hu, Hwai-Tsu
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/84g66d

碩士 === 國立宜蘭大學 === 電子工程學系碩士班 === 105 === In the digital era, multimedia data appear everywhere in life. However, the easy manner by which digitized data can be stored, copied, and distributed, has led to numerous copyright infringement problems. To solve this issue, "digital watermarking" has been recently proposed as a copyright protection method, whereby proprietary information can be embedded into multimedia, and later extracted to prove ownership. This technology has the potential to settle copyright disputes and attain copyright protection. This thesis proposes combining several techniques including discrete wavelet transform (DWT), time-spread echo (TSE), rational dither modulation (RDM) and log coordinate mapping (LCM) to achieve effective blind audio watermarking. To verify whether the time scale of a watermarked audio signal has been altered, we deliberately incorporate TSE into a competent watermarking framework called DWT-RDM. In case the playback speed was modified, the original time scale can be restored according to the gauge of TSE. In addition, the locations for watermark embedding and extraction can be aligned to enable correct watermark retrieval. We discuss the most suitable arrangement for deploying TSE inside the DWT-RDM algorithm. All efforts are meant to obtain a balance between robustness and imperceptibility for audio watermarks. In the premise that the watermark robustness shall not be adversely degraded, we verify the adequate time duration for acquiring quantization steps. Both signal-to-noise ratio (SNR) and objective difference grade (ODG) can be further improved by choosing suitable DWT subbands for watermarking and by progressively adjusting the embedding strength. Experiment results show that the embedded TSE hardly affects the resulting audio quality. The TSE actually empowers the primordial DWT-RDM algorithm to resist playback speed modification.