An audio watermark embedding and extraction method based on low-frequency carrier

Abstract

The present invention relates to an audio watermark embedding and extraction method based on a low-frequency carrier wave, and belongs to the field of audio processing technology. First, a binary watermark image is converted into a one-dimensional array, and then an XOR operation is performed with a key of the same length to improve the security of the digital audio watermark. Secondly, a sine wave signal with the same number of sample points is generated according to the number of sample points of the original audio signal, and then the sign of the sample value of each half cycle of the sine wave is transformed according to the watermark information, and finally the transformed signal is superimposed with the original audio signal to obtain an audio signal containing the watermark information. When extracting the watermark, the audio signal containing the watermark is Fourier transformed, and the sample points with a frequency greater than 5 times the sine wave frequency after the Fourier transform are set to zero, thereby obtaining a signal containing the watermark information, and the watermark information is determined by judging the sign of each half cycle of the signal, and finally the original key is XORed and reconstructed to obtain the extracted watermark information.

Description

Technical Field

[0001] The invention relates to an audio watermark embedding and extraction method based on a low-frequency carrier wave, and belongs to the technical field of audio processing. Background Art

[0002] With the continuous development of network technology, audio copyright and authentication issues have received more attention. Therefore, digital audio watermarking has been widely used in the fields of copyright protection of music and voice files.

[0003] Most of the existing audio watermarking technologies embed watermarks in the audio transform domain, which is relatively complicated. Therefore, the present invention improves the traditional digital audio watermarking algorithm, embeds watermark information in the time domain, and uses a low-frequency sinusoidal wave signal as a watermark carrier, thereby improving the imperceptibility of the watermark information and having good robustness against noise interference. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide an audio watermark embedding and extraction method based on a low-frequency carrier wave, thereby improving the imperceptibility of the audio watermark and solving the above-mentioned problem.

[0005] The technical solution of the present invention is: an audio watermark embedding and extraction method based on low-frequency carrier, the specific steps are:

[0006] Step 1: Watermark information preprocessing:

[0007] The matrix of binary watermark information (size is m*m) is placed after each column of data in the previous column, and processed in sequence to obtain a column of one-dimensional binary vectors with a length of n, which is then XORed with a column of keys of the same length consisting of 0 and 1 to obtain the watermark information to be embedded.

[0008] Step 2: Embed the watermark into the original audio:

[0009] A sine wave signal of a specific frequency is generated according to the original audio and the watermark information to be embedded; the sign of the sine wave signal is transformed according to the watermark information to be embedded, so as to obtain a waveform signal with the watermark information; the waveform signal with the watermark information is added to the sampling value of the original audio signal, thereby obtaining an audio signal containing the watermark information.

[0010] Step 3: Extract watermark from audio:

[0011] After performing Fourier transform on the audio signal containing watermark information, set the data of the sampling points whose frequencies are greater than 5 times the sine wave frequency to zero, and then perform inverse Fourier transform on it to obtain the signal containing watermark information; perform watermark extraction on the signal containing watermark information, perform exclusive OR operation on the extracted one-dimensional vector and the secret key, and finally obtain the watermark information through matrix reconstruction.

[0012] The specific steps of the said Step2 are as follows:

[0013] Step2.1: Determine the number of sampling points of the sine wave to be N according to the number of sampling points N of the original audio.

[0014] Step2.2: The time t of the audio signal is (N - 1) / Fs, where Fs is the sampling frequency of the audio.

[0015] Step2.3: According to the length n of the watermark information to be embedded described in Step1, calculate the sine wave frequency estimate f' = n / 2t, and ensure that the sine wave frequency satisfies f' < f < 20, and the phase of the sine wave is 0.

[0016] Step2.4: Transform the said sine signal according to the rule that 2 bits of watermark information are represented in one period.

[0017] Step2.5: The transformation method is: when the watermark information is 1, change the signs of the sampling points in the corresponding half period to positive signs, and when the watermark information is 0, change the signs to negative signs, so as to obtain the signal with watermark information.

[0018] Step2.6: Add the signal with watermark information to the sampling values of the original audio signal to obtain the audio signal containing watermark information.

[0019] The specific steps of the said Step3 are as follows:

[0020] Step3.1: Perform Fourier transform on the audio signal containing watermark information.

[0021] Step3.2: Calculate the corresponding frequency of each sampling point, then set the data of the sampling points whose frequencies are greater than 5 times the sine wave frequency f to zero, and then perform inverse Fourier transform to obtain the signal containing watermark information.

[0022] Step3.3: Extract the said signal containing watermark information according to the following rule: when the sampling value of the signal is greater than 0 at the median of each half period of the sine wave signal, it is determined to be 1, otherwise it is determined to be 0, so as to obtain a group of one-dimensional vectors.

[0023] Step3.4: Perform exclusive OR operation on the said one-dimensional vector and the original secret key, and take every m bits of the calculated one-dimensional vector in order as a column to obtain a matrix of size m*m, which is the extracted watermark information.

[0024] The beneficial effects of the present invention are as follows: the present invention uses low-frequency sine waves as the carrier of watermark information, so that the audio embedded with the watermark is highly imperceptible to the human ear, and at the same time makes the watermark embedding and extraction calculation simple and easy to implement. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a flow chart of the steps of the present invention;

[0026] Figure 2 It is an experimental process diagram of the present invention;

[0027] Figure 3 It is the original watermark and the extracted watermark image of the present invention. DETAILED DESCRIPTION

[0028] The present invention will be further described below in conjunction with the accompanying drawings and specific implementation methods.

[0029] like Figure 1 As shown in the figure, an audio watermark embedding and extraction method based on low-frequency carrier is used. When embedding the watermark, the binary watermark image is first converted into a one-dimensional array, and then an XOR operation is performed with a key of the same length to improve the security of the digital audio watermark. Secondly, a sine wave signal with the same number of sample points is generated according to the number of sample points of the original audio signal. The amplitude of the sine wave is determined by the amplitude of the audio signal, the phase is 0, and the frequency value is determined by the length of the watermark to be embedded. At the same time, the frequency is ensured to be lower than 20Hz. Then, the sign of the sample value of each half cycle of the sine wave is transformed according to the watermark information, and the transformed signal is superimposed with the original audio signal to obtain an audio signal containing the watermark information. When extracting the watermark, the audio signal containing the watermark is Fourier transformed, and the sample points with a frequency greater than 5 times the frequency of the sine wave after Fourier transformation are set to zero, thereby obtaining a signal containing the watermark information. The watermark information is determined by judging the sign of each half cycle of the signal. Finally, the original key is used to perform an XOR operation on it and reconstruct it to obtain the extracted watermark information.

[0030] The specific steps are:

[0031] Select watermark as Figure 3 As shown in (1), the size is 15*15 pixels.

[0032] The watermark matrix is ​​processed sequentially by placing each column of data after the previous column of data to generate a one-dimensional binary vector of length n=225, which is then XORed with a binary key of the same length to obtain the watermark information a(i) to be embedded.

[0033] In this embodiment, a music signal is selected as the original audio x, such as Figure 2As shown in (a), the sampling frequency Fs = 44100 Hz, and the number of sampling points N = 524288.

[0034] The frequency of the sine wave is calculated by the formula f' = (n*Fs) / (2*(N-1)), and f' is about 9.47Hz. Take the sine wave frequency f = 10Hz, which satisfies f' <f<20。

[0035] The obtained sinusoidal wave signal y is transformed according to the following formula, and one cycle represents 2 bits of watermark information.

[0036]

[0037] The number of sampling points in a half cycle k=Fs / (2*f).

[0038] Add the watermarked signal y to the sampled value of the original audio signal x to obtain the audio signal containing the watermark information. Figure 2 As shown in (b).

[0039] When extracting the watermark, the audio signal containing the watermark information is subjected to Fourier transform.

[0040] Calculate the frequency fn=n / (2*N*Fs) corresponding to each sampling point, then set the sampling values ​​with a frequency greater than 50Hz to zero, and then perform inverse Fourier transform to obtain a signal containing watermark information, such as Figure 2 As shown in (d).

[0041] The signal containing the watermark information is extracted according to the following rules:

[0042] When the median of the signal sampling value in each half cycle of the sine wave signal is greater than 0, it is judged to be 1, otherwise it is judged to be 0, thereby obtaining a set of one-dimensional vectors b.

[0043] The one-dimensional vector is XORed with the original key, and the calculated one-dimensional vector is sequentially used as a column of a matrix for every 15 bits to obtain the extracted watermark information as follows: Figure 3 As shown in (2).

[0044] The watermarked audio obtained by the present invention is easy to extract and detect due to its low embedding frequency, and the calculation amount of the method is smaller than that of watermark embedding in the transform domain, thereby reducing the cost.

[0045] The specific implementation modes of the present invention are described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above implementation modes, and various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present invention.

Claims

1. An audio watermark embedding and extraction method based on low-frequency carrier, characterized in that: Step1: Process the matrix of binary watermark information in sequence according to the rule that each column of data is placed after the previous column of data, to obtain a one-dimensional binary vector, and then perform an exclusive OR operation with a binary key of the same length, so as to obtain the watermark information to be embedded; Step2: Generate a sine wave signal with a specific frequency according to the original audio and the watermark information to be embedded, and perform a sign transformation on the sine wave signal according to the watermark information to be embedded, so as to obtain a waveform signal with watermark information, and add the sampling value of the waveform signal with watermark information to the sampling value of the original audio signal, thereby obtaining an audio signal containing watermark information; The specific steps of Step2 are as follows: Step2.1: Determine that the sampling points of the sine wave are N according to the number of sampling points N of the original audio; Step2.2: The time t of the audio signal = (N - 1) / Fs, where Fs is the sampling frequency of the audio; Step2.3: According to the length n of the watermark information to be embedded described in Step1, calculate the sine wave frequency estimate f’ = n / 2t, and ensure that the sine wave frequency satisfies f’ < f < 20, and the phase of the sine wave is 0; Step2.4: Transform the sine wave signal according to the rule that one period represents 2 bits of watermark information; Step2.5: The transformation method is: when the watermark information is 1, change the sign of the sampling points in the corresponding half period to a positive sign, and when the watermark information is 0, change the sign to a negative sign, so as to obtain a signal with watermark information; Step2.6: Add the signal with watermark information to the sampling value of the original audio signal to obtain an audio signal containing watermark information; Step3: Perform a Fourier transform on the audio signal containing watermark information, set the sampling point data with a frequency greater than 5 times the sine wave frequency to zero after the transformation, and then perform an inverse Fourier transform to obtain a signal containing watermark information; Perform watermark extraction on the signal containing watermark information, perform an exclusive OR operation on the extracted one-dimensional vector and the original key, and finally reconstruct the matrix to obtain the extracted watermark information.

2. The audio watermark embedding and extraction method based on low-frequency carrier according to claim 1 is characterized in that: The specific steps of Step3 are as follows: Step3.1: Perform a Fourier transform on the obtained audio signal containing watermark information, calculate the frequency magnitude corresponding to each sample point, set the sample points with a frequency greater than 5 times the sine wave frequency to zero, and then perform an inverse Fourier transform to obtain a signal containing watermark information; Step3.2: Extract the signal containing watermark information according to the following rule: when the sign of the median value of each half period of the sine wave signal is positive, it is determined to be 1, otherwise it is 0, so as to obtain a group of one-dimensional vectors, and then perform an exclusive OR operation with the original key, and reconstruct the array to obtain the extracted watermark information.

Images