Scholarlink

<p>Digital watermarking is a widely used technique for protecting copyright information from unauthorized users. This paper proposes a robust and imperceptible audio watermarking to safeguard the secret images using 2D Fast Fourier Transform and Framelet Transform, which enables perfect reconstruction of both the original signal and the embedded secret image with zero Bit Error Rate(BER). The proposed technique embeds the binary image into singular values of audio signal, to ensure secure and reliable watermarking. The secret image is divided into small square images which it is scrambled using Arnold Transformation (AT) and then embedded in a random sequence using Logistic map to strengthen the security. A quality-constrained multi-metrics based optimization is proposed to determine the optimal scaling factor. The Perceptual Evaluation of Speech Quality (speech) and Objective Difference Grade scores(music) are taken as perceptual quality constraints, while maintaining the Signal-to-Noise Ratio (SNR) of greater than 20 dB with zero BER (no-attack) conditions based optimization algorithm is proposed to enhance the imperceptibility and robustness. The performance of proposed technique is evaluated for payload of 2250 bps for music, 2068 bps for audio and 1500 bps for speech and noisy datasets. Experimental results gives average SNR, PESQ and ODG scores of 58.60dB, 3.97 and −0.2, respectively. Moreover, it gives SNR of 30 dB for real surround sound dataset. The proposed technique effectively balances the imperceptibility, robustness and payload trade-off against stirmark, common signal processing, desynchronization and real world codec/compression attacks with low computational complexity compared to existing techniques, while ensuring better performance across diverse audio signals.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Robust Audio Watermarking using hybrid FFT-FLT-AT Technique with High Imperceptibility

  • 期刊论文

  • 发表日期：

    2026年7月27日

查看全文
  • Sumitra S,
  • Aniruddha Kanhe

摘要

Digital watermarking is a widely used technique for protecting copyright information from unauthorized users. This paper proposes a robust and imperceptible audio watermarking to safeguard the secret images using 2D Fast Fourier Transform and Framelet Transform, which enables perfect reconstruction of both the original signal and the embedded secret image with zero Bit Error Rate(BER). The proposed technique embeds the binary image into singular values of audio signal, to ensure secure and reliable watermarking. The secret image is divided into small square images which it is scrambled using Arnold Transformation (AT) and then embedded in a random sequence using Logistic map to strengthen the security. A quality-constrained multi-metrics based optimization is proposed to determine the optimal scaling factor. The Perceptual Evaluation of Speech Quality (speech) and Objective Difference Grade scores(music) are taken as perceptual quality constraints, while maintaining the Signal-to-Noise Ratio (SNR) of greater than 20 dB with zero BER (no-attack) conditions based optimization algorithm is proposed to enhance the imperceptibility and robustness. The performance of proposed technique is evaluated for payload of 2250 bps for music, 2068 bps for audio and 1500 bps for speech and noisy datasets. Experimental results gives average SNR, PESQ and ODG scores of 58.60dB, 3.97 and −0.2, respectively. Moreover, it gives SNR of 30 dB for real surround sound dataset. The proposed technique effectively balances the imperceptibility, robustness and payload trade-off against stirmark, common signal processing, desynchronization and real world codec/compression attacks with low computational complexity compared to existing techniques, while ensuring better performance across diverse audio signals.