An Adaptive Ultrasonic Image Artifact Removal Method Based on SoS Function Modulation

Abstract

The present invention discloses an adaptive ultrasonic image artifact removal method based on SoS function modulation. Using the ultrasonic echo signal measured during conventional ultrasonic detection, it is modulated with the SoS function to obtain a transform domain signal. Then, the modulated transform domain signal is sampled at a low rate at equal intervals to obtain the discrete sequence values of the modulation signal. Next, the discrete Fourier transform is performed on the discrete sequence to obtain the Fourier coefficient sequence of the original ultrasonic echo signal. The spectral estimation algorithm is used to estimate the parameters of the ultrasonic echo signal, obtain the characteristic parameters of the ultrasonic detection signal, and combine with the ultrasonic sensor waveform to perform adaptive waveform reconstruction of the echo signal. Using the reconstructed signal for acoustic imaging, an artifact-free ultrasonic pseudo-color image is obtained. The present invention effectively removes the artifacts of the ultrasonic image and retains information such as the position, shape, and size of the defects, is applicable to industrial ultrasonic imaging, and has good application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of non-destructive testing, and particularly to an adaptive ultrasonic image artifact removal method based on SoS function modulation. Background Art

[0002] Industrial ultrasonic testing uses an ultrasonic beam to scan a test piece, receives the echo signal of the target body, and uses an ultrasonic imaging algorithm to visualize the defect echo information of the test piece to form an ultrasonic image. The position, size, shape and other information of the defect can be intuitively displayed in the image, enabling the inspector to observe the defects of the test piece from the image and providing a basis for defect detection and evaluation. Artifacts in industrial ultrasonic images refer to non-target images formed in non-target areas, which are caused by principle errors and random errors. The existence of artifacts will affect the accurate judgment of actual defects.

[0003] There are mainly three reasons for the generation of artifacts. The first is caused by the algorithm for solving the spatial sound field during imaging, which belongs to the category of principle error generation. The artifacts caused by this reason are sometimes also called artifacts caused by equal sound path lines. The second is caused by the tailing phenomenon of ultrasonic echo signals, which belongs to the category of random error generation. The after-vibration of the received echo signal lengthens the duration of the echo. These tailing signals have non-zero amplitudes and are misinterpreted as useful echo signals when solving the sound field. The third is the artifacts caused by random noise, which has a random distribution characteristic and makes the image blurred. Although many methods have been applied to the removal of artifacts in ultrasonic images and certain artifact removal effects have been achieved, the artifacts caused by these three reasons have not been completely solved. For example, by discriminating the effective equal sound path lines in the sound field solution, the artifacts caused by equal sound path lines can be effectively removed; using sensors or receiving circuits with high damping can weaken the tailing phenomenon of echo signals; using filtering methods can filter out some echo noises, but the artifact problem has always been a common problem in the field of ultrasonic imaging, and there is no method that can effectively remove these artifacts.

[0004] Aiming at the artifact problem in ultrasonic images, the present invention proposes an adaptive ultrasonic image artifact removal method based on SoS function modulation. The ultrasonic echo signal collected by sensor detection and processing is modulated by the SoS function, and discrete signal sampling sequence values are obtained by equidistant sampling. Then, the discrete Fourier transform of the discrete signal sampling sequence is performed through calculation to obtain the Fourier coefficient information of the original sound echo signal. The signal parameter estimation is realized by using the spectral estimation algorithm, the original signal is adaptively reconstructed according to the Gaussian pulse function, and finally acoustic imaging is performed on it to obtain a defect ultrasonic image with artifacts removed. Summary of the Invention

[0005] The purpose of the present invention is to provide an adaptive ultrasonic image artifact removal method based on SoS function modulation, which can correct the signal detected by the ultrasonic sensor when a test piece defect is detected, and obtain an ultrasonic image with artifacts removed through an imaging algorithm. This method retains the defect information of the ultrasonic image and has a significant artifact removal effect, which can improve the accuracy and efficiency of defect assessment. To achieve the above-mentioned invention objectives, the technical solution provided by the present invention is as follows: An adaptive ultrasonic image artifact removal method based on SoS function modulation, comprising the following steps:

[0006] 1) Using an ultrasonic sensor to detect the test piece, receive the echo signal, amplify, filter, and extract the envelope of the echo signal to obtain the ultrasonic echo signal x(t) obtained in each scan;

[0007] 2) Using the SoS function to perform amplitude modulation on the ultrasonic echo signal x(t) to obtain the transform domain signal y(t);

[0008] 3) sampling the transform domain signal y(t) at equal intervals to obtain a discrete sparse sampling sequence y[n] of the transform domain signal;

[0009] 4) Performing discrete Fourier transform on the discrete sparse sampling sequence y[n], whose Fourier coefficients contain information of the ultrasonic echo signal;

[0010] 5) Apply the spectrum estimation algorithm to estimate the parameters of the Fourier coefficients to obtain the delay and amplitude parameters of the ultrasonic echo signal, and reconstruct the ultrasonic echo signal waveform in combination with the ultrasonic sensor transmission waveform;

[0011] 6) Using the reconstructed ultrasonic echo signal to perform acoustic imaging, a pseudo-color ultrasonic image with artifacts removed is obtained.

[0012] Furthermore, in the above step 1), the ultrasonic echo signal is generated by the high-voltage pulse exciting the piezoelectric chip in the ultrasonic transducer to oscillate. Assume that the Gaussian model of the ultrasonic echo signal x(t) is represented by r(t), and the mathematical model of r(t) is:

[0013]

[0014] in, E represents the amplitude of the excitation pulse; η is its pulse width factor; f0 is the center frequency of the chip; K represents the number of echoes; φ k is the initial phase; parameter t k and a k Carrying the defect echo information of the test block to be inspected, the arrival time t k Reflects the position information of the reflecting surface, echo amplitude a k Reflects the loss of ultrasonic sound intensity.

[0015] Furthermore, the SoS function in the above step 2) is realized by weighted sum of sinc functions, and its frequency domain expression is:

[0016]

[0017] Where τ is the duration of the signal; ω is the signal frequency; Ψ represents a continuous set of integers, the length of which is determined by the information freedom of the signal; c m Represents the weighting coefficient, satisfying

[0018] Converting the SoS function to the time domain, its expression is:

[0019]

[0020] The waveform of s(t) is given by {c m} m∈Ψ Changing these parameters will affect the time-frequency domain response of the SoS function and achieve different modulation effects.

[0021] Furthermore, in the above step 2), the ultrasonic echo signal x(t) is amplitude modulated using the SoS function to obtain the modulated transform domain signal y(t).

[0022]

[0023] Where τ is the duration of the signal.

[0024] Furthermore, in the above step 3), the transform domain signal y(t) is sampled at equal intervals, and the sampling frequency is determined according to the innovation rate of the signal, and the sampling frequency is not less than the innovation rate of the signal, and a discrete sparse sampling sequence y[n] is obtained after sampling;

[0025] The new interest rate is obtained in the following way:

[0026] Assuming that the ultrasonic echo signal x(t) satisfies the condition that it can be represented by a finite number of degrees of freedom of information, it can be expressed in a way based on the degrees of freedom of information, and its expression is,

[0027]

[0028] Where c n,r and t n Represent the amplitude parameter and delay parameter of the signal respectively, is the basis function, which can be selected as an impulse function or a Gaussian function; from the expression of the signal x(t) based on the finite information rate, it can be seen that the signal is only composed of c n,r and t n OK. Use the continuous function C xLet (t1,t2) represent the information freedom degree of x(t) within the time interval [t1,t2], then the signal can be represented by a finite number of free parameters. Calculate the innovation rate based on the local time domain interval of the signal, that is, the finite information freedom degree, and the signal can be truncated by a time window to obtain the local innovation rate P of the signal. τ (t),

[0029]

[0030] The local innovation rate characterizes the number of information freedom degrees of the signal within the duration interval τ.

[0031] Furthermore, in step 4) above, the expression of x(t) based on the finite information freedom degree is further simplified to form a form u(t) represented by a Dirac flow signal, which is expressed as

[0032]

[0033] Among them, t k ∈[0,τ), a k ∈R, t k and a k are the echo signal delay and amplitude in the Gaussian model of the ultrasonic echo signal x(t), and are also the main characteristic parameters of the echo signal to be estimated subsequently;

[0034] Expand the signal u(t) in Fourier series, and its Fourier coefficient U[m] can be expressed in the form of a weighted sum of power series, that is,

[0035]

[0036] The Fourier coefficient U[m] contains the delay and amplitude information of the ultrasonic echo signal x(t), that is, the Fourier coefficient information of the original ultrasonic echo signal.

[0037] Furthermore, step 5) above specifically includes the following steps:

[0038] 5.1) Solve the weighted value and exponent of the power series through a spectral estimation algorithm;

[0039] 5.2) Convolve the filter coefficient h m with C[m], C[m] = τU[m]. If the Z-transform of the nulling filter satisfies H(q k ) = 0, then the result of their convolution is zero, and the filter coefficient h m that meets the requirements can be obtained, which is expressed as:

[0040]

[0041] Let h0 = 1, and expand it to get:

[0042] h1C[m - 1] + h2C[m - 2] + … + h K C[m - K] = -C[m]

[0043] The above equation contains K unknowns {h1, h2, …, h K}, and at least K equations are required to solve it, that is, at least 2K consecutive Fourier series coefficients are required for the system of equations to have a unique solution.

[0044] 5.3) Substitute the Fourier coefficients and the number of signal pulses into the system of equations in 5.2) to obtain the information of the parameters to be estimated

[0045] Let a continuous integer interval of length L be set, and Λ represents the value range of the Fourier series coefficients of the signal. Among them, Λ = [-l, …, 0, …, l], l is the number of pulses in τ, and L = 2l + 1. Then, the relationship between the number of Fourier coefficients L and the number of signal pulses K needs to satisfy:

[0046] L = 2l + 1 ≥ 2K

[0047] It can be seen from the above derivation that only by obtaining 2K consecutive Fourier coefficients can the amplitude and delay parameters of the signal be estimated by the above method

[0048] 5.4) According to the estimated parameters Combined with the Gaussian pulse model r(t) of the ultrasonic echo signal, adaptively reconstruct the ultrasonic echo signal that retains the important information of x(t).

[0049] Furthermore, in the above step 5.1), the spectral estimation algorithm uses the annihilating filter method to solve the algorithm, including the following steps:

[0050] 5.1.1) Construct an annihilating filter with coefficients Let where m ∈ z, and the z-transform of the annihilating filter is

[0051]

[0052] Let the zeros of H(z) be Let h0 = 1, and after factorization, H(z) can be expressed as

[0053]

[0054] When all the delay parameters are distinct, the zeros of the filter can uniquely represent the pulse delay parameters of the signal;

[0055] 5.1.2) Find the coefficients h of the annihilating filterm , by finding the roots \(q\) of \(H(z)\) k , the time delay parameter \(t\) is obtained k ;

[0056] 5.1.3) Use \(U[m]\) to solve for the amplitude parameter \(a\) k for solution.

[0057] Furthermore, the sound field of the measured area is calculated for the reconstructed signal in step 6) above, the sound field intensities at spatial points are synthesized, and the magnitude of the synthesized sound field intensity is represented by color gradations, thereby obtaining an ultrasonic pseudo-color image with artifacts removed.

[0058] The present invention has the following beneficial effects:

[0059] An adaptive ultrasonic image artifact removal method based on SoS function modulation is invented, which can correct the signal data of the defects detected by an ultrasonic sensor, and obtain an ultrasonic image without artifacts through an imaging algorithm. Starting from the data, this algorithm modulates the signal using the SoS function, obtains the Fourier coefficient information of the original ultrasonic echo signal through arithmetic processing, uses the spectral estimation algorithm to realize signal parameter estimation and reconstruction, corrects the artifact data of the data, and only retains the effective information of the defect for imaging. By comparing the information such as amplitude and time delay between the reconstructed signal and the original ultrasonic echo signal, and the final actual imaging effect, the effectiveness of the artifact removal algorithm is illustrated. On the basis of retaining the defect information of the ultrasonic image, this algorithm has a remarkable artifact removal effect, and can improve the accuracy and efficiency of defect evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a flowchart of the method of the present invention;

[0061] Figure 2 is the time-domain and frequency-domain waveform diagrams of the SoS function of the present invention; wherein, (a) time domain; (b) frequency domain;

[0062] Figure 3 is Embodiment 1 of the present invention; wherein, (a) schematic diagram of the through-hole defect distribution; (b) original signal and reconstructed signal of the through-hole defect; (c) ultrasonic image of the through-hole defect with artifacts; (d) ultrasonic image of the through-hole defect with artifacts removed by using the artifact removal method of the present invention;

[0063] Figure 4 is Embodiment 2 of the present invention; wherein, (a) schematic diagram of the straight groove defect distribution; (b) original signal and reconstructed signal of the straight groove defect; (c) ultrasonic image of the straight groove defect with artifacts; (d) ultrasonic image of the straight groove defect with artifacts removed by using the artifact removal method of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0064] The specific form of the present invention will be further described below in conjunction with the accompanying drawings and embodiments, and its process is as follows Figure 1 It should be noted that the present invention can also be applied through other equivalent implementation manners. The implementation manners described in the following embodiments illustrate the process and effects of the present invention by way of example.

[0065] As Figure 1 shown, the present invention is an adaptive ultrasonic image artifact removal method based on SoS function modulation, including the following steps:

[0066] 1) Use an ultrasonic sensor to detect the test piece, receive the echo signal, and perform amplification, filtering, and envelope extraction on the echo signal to obtain the ultrasonic echo signal x(t) obtained from each scan; As a preferred embodiment of the present invention, the sensor performs mechanical scanning in a plane at equal step lengths during detection.

[0067] 2) Use the SoS function to perform amplitude modulation on the ultrasonic echo signal x(t) to obtain the transform domain signal y(t);

[0068] 3) Sample the transform domain signal y(t) at equal intervals to obtain the discrete sparse sampling sequence y[n] of the transform domain signal;

[0069] 4) Perform discrete Fourier transform on the discrete sparse sampling sequence y[n], and its Fourier coefficients contain the information of the ultrasonic echo signal;

[0070] 5) Apply a spectral estimation algorithm to estimate the parameters of the Fourier coefficients, obtain the time delay and amplitude parameters of the ultrasonic echo signal, and reconstruct the ultrasonic echo signal waveform in combination with the ultrasonic sensor emission waveform;

[0071] 6) Use the reconstructed ultrasonic echo signal for acoustic imaging to obtain an ultrasonic pseudo-color image with artifacts removed.

[0072] As a preferred embodiment of the present invention, in step 1) above, the ultrasonic echo signal is generated by exciting the piezoelectric wafer in the ultrasonic transducer with a high-voltage pulse. Let the Gaussian model of the ultrasonic echo signal x(t) be represented as r(t), and the mathematical model of r(t) is

[0073]

[0074] where E represents the amplitude of the excitation pulse; η is its pulse width factor; f0 is the center frequency of the wafer; K represents the number of echoes; φ k is the initial phase; the parameters t k and a k carry the defect echo information of the test block to be detected. The arrival time t k reflects the position information of the reflecting surface, and the echo amplitude a kReflect the sound intensity loss of ultrasonic waves.

[0075] As a preferred embodiment of the present invention, in the above step 2), the SoS function is implemented by the weighted sum of the sinc function, and its frequency-domain expression is:

[0076]

[0077] where τ is the signal duration length; ω is the signal frequency; Ψ represents a set of consecutive integers, and the set length is determined by the information degrees of freedom of the signal; c m represents the weighting coefficient, satisfying

[0078] Converting the SoS function to the time domain, its expression is,

[0079]

[0080] The waveform of s(t) is determined by {c m}, m∈Ψ Changing this series of parameters will affect the time-frequency domain response of the SoS function and can achieve different modulation effects. Schematic diagrams of the time-domain and frequency-domain waveforms of the SoS function are shown in Figure 2 (a) and (b) respectively, where the frequency in the figure

[0081] As a preferred embodiment of the present invention, in the above step 2), the ultrasonic echo signal x(t) is amplitude-modulated by the SoS function to obtain the modulated transform-domain signal y(t),

[0082]

[0083] where τ is the signal duration length.

[0084] As a preferred embodiment of the present invention, in the above step 3), the transform-domain signal y(t) is sampled at equal intervals, and the sampling frequency is determined according to the innovation rate of the signal. The sampling frequency is not less than the innovation rate of the signal. After sampling, a discrete sparse sampling sequence y[n] is obtained;

[0085] The innovation rate is obtained in the following manner:

[0086] Assume that the ultrasonic echo signal x(t) satisfies the condition that it can be characterized by a finite number of information degrees of freedom, then it can be expressed in a way based on the information degrees of freedom, and its expression is,

[0087]

[0088] In the formula, c n,r and t n represent the amplitude parameter and time-delay parameter of the signal, respectively, As the basis function, it can be selected as the impulse function or the Gaussian function; from the expression of the signal x(t) based on the finite information rate, it can be seen that this signal is only determined by c n,r and t n Determined. Use the continuous function C x (t1, t2) to represent the information freedom degree of x(t) within the time interval [t1, t2], then the signal can be represented by a finite number of free parameters. Calculate the innovation rate with the local time domain interval of the signal, that is, the finite information freedom degree, and the signal can be truncated through a time window to obtain the local innovation rate P τ (t),

[0089]

[0090] The local innovation rate characterizes the number of information freedom degrees of the signal within the duration interval τ.

[0091] As a preferred embodiment of the present invention, in the above step 4), the expression of x(t) based on the finite information freedom degree is further simplified to form a form u(t) represented by the Dirac flow signal, expressed as,

[0092]

[0093] Among them, t k ∈[0, τ), a k ∈R, t k and a k Are the echo signal delay and amplitude in the Gaussian model of the ultrasonic echo signal x(t), and are also the main characteristic parameters of the echo signal to be estimated subsequently;

[0094] Expand the signal u(t) in Fourier series, and its Fourier coefficient U[m] can be expressed in the form of a weighted sum of power series, that is,

[0095]

[0096] The Fourier coefficient U[m] contains the delay and amplitude information of the ultrasonic echo signal x(t), that is, the Fourier coefficient information of the original ultrasonic echo signal.

[0097] As a preferred embodiment of the present invention, the above step 5) specifically includes the following steps:

[0098] 5.1) Solve the weighted value of the power series and the exponent through the spectral estimation algorithm;

[0099] 5.2) Convolve the filter coefficient h m with C[m], C[m] = τU[m], if the Z-transform of the nulling filter satisfies H(q k) = 0, then the result of their convolution is zero, and the filter coefficients h that meet the requirements can be obtained m , which is expressed as:

[0100]

[0101] Let h0 = 1, and expanding it gives:

[0102] h1C[m - 1] + h2C[m - 2] + … + h K C[m - K] = -C[m]

[0103] The above equation contains K unknowns {h1, h2, …, h K}}, and at least K equations are required to solve, that is, at least 2K consecutive Fourier series coefficients are required for the system of equations to have a unique solution.

[0104] 5.3) Substitute the Fourier coefficients and the number of signal pulses into the system of equations in 5.2) to obtain the information of the parameters to be estimated

[0105] Let there be a continuous integer interval of length L, and Λ represents the value interval of the signal Fourier series coefficients, where Λ = [-l, …, 0, …, l], l is the number of pulses in τ, and L = 2l + 1. Then, the relationship between the number of Fourier coefficients L and the number of signal pulses K needs to satisfy:

[0106] L = 2l + 1 ≥ 2K

[0107] From the above derivation, it can be seen that by only obtaining 2K consecutive Fourier coefficients, the amplitude and time delay parameters of the signal can be estimated using the above method

[0108] 5.4) According to the estimated parameters Combined with the Gaussian pulse model r(t) of the ultrasonic echo signal, adaptively reconstruct the ultrasonic echo signal that retains the important information of x(t).

[0109] As a preferred embodiment of the present invention, in the above step 5.1), the spectral estimation algorithm uses the annihilating filter method to solve the algorithm, including the following steps:

[0110] 5.1.1) Construct an annihilating filter with coefficients , let where m ∈ Z, and the z-transform of the annihilating filter is

[0111]

[0112] Let the zeros of H(z) be Let h0 = 1, and after factorization, H(z) can be expressed as

[0113]

[0114] When all the time-delay parameters are different from each other, the zeros of the filter can uniquely represent the impulse time-delay parameters of the signal;

[0115] 5.1.2) Calculate the coefficients h of the nulling filter m , by finding the roots q of H(z) k , and obtaining the time-delay parameter t k ;

[0116] 5.1.3) Use U[m] to solve for the amplitude parameter a k for solution.

[0117] As a preferred embodiment of the present invention, the reconstructed signal in the above step 6) is used to calculate the sound field in the area to be measured, the sound field intensities at spatial points are synthesized, and the magnitude of the synthesized sound field intensity is represented by a color scale, thereby obtaining an ultrasonic pseudo-color image with artifacts removed.

[0118] The technical solution of the present invention will be further described below in conjunction with specific embodiments. Specific Embodiment 1

[0120] (1) Select a standard defect test block of through-hole type for testing. The test block and the geometric distribution of the defects are as Figure 3 (a) shown. Set the diameter of the through-hole defects to be 2 mm uniformly. The ultrasonic sensor uses a water-immersion point-focusing probe. The center frequency of the sensor is 5 MHz, the diameter of the probe is 13.0 mm, the model is I2-1P25F70-H (IGI1320), and the focal length is 78 mm (the focal length in water). The test block is made of aluminum material, and its ultrasonic sound speed is 6300 m / s. Adopt a vertical incidence detection method. The couplant is water, and the wave speed in the couplant is 1480 m / s. Set the focus at the center position of the test block, that is, the distance from the focus in the workpiece to the workpiece surface is 17 mm, then the water layer thickness is 20 mm. During detection, use the sensor's self-excitation and self-reception mode. The pulse width is 200 ns, the voltage is 200 V. Control the probe to move at an equal step size with a step size of 0.1 mm through a mechanical structure, and the moving speed is 10 mm / s. Take the upper left corner of the test block as the starting origin and detect along the X-axis direction in the XOY plane, so that the scanning range of the sensor covers the entire test block to be tested. Amplify (the gain is 25 dB), filter (a band-pass filter of 2 - 6 MHz), and envelope extraction are performed on the echo signals received by the sensor. Receive the echo and obtain the ultrasonic echo signals obtained from each detection. The sampling frequency is 80 MHz. Since the ultrasonic signal is continuously attenuated, select the data of one surface wave and one bottom wave for processing and imaging.

[0121] (2) Implement the SoS function through sinc function weighting and summation. The weighting coefficient c mIf all are taken as 1, the time-domain and frequency-domain waveform diagrams of the SoS function are as follows Figure 2 shown. The collected ultrasonic echo signal is modulated by the SoS function, and the signal is transformed in the time domain so that the transform-domain information required for parameter estimation can be obtained from the downsampled values during the subsequent reconstruction process.

[0122] (3) The modulated signal is sampled at equal intervals with a time interval of 0.067 μs. It is required that the sampling frequency is not less than the innovation rate of the signal, and the discrete signal sampling sequence values are obtained;

[0123] (4) The discrete Fourier transform of the discrete signal sampling sequence is performed through operations to obtain the Fourier coefficient information of the original ultrasonic echo scan signal;

[0124] (5) The signal amplitude and time-delay parameter estimation is realized by using the spectral estimation algorithm, and the waveform recovery is carried out by combining the known waveform with Gaussian characteristics to reconstruct the signal. The reconstruction accuracy is 13.5, and the number of echo signals of the signal is 2, that is, the number of signal pulses. The waveforms of the primary surface wave and the defect are reconstructed by using the amplitude and time delay where the surface wave and the defect are located. The reconstructed signal and the original are as Figure 3 (b) shown.

[0125] (6) The ultrasonic echo signal before being processed by the ultrasonic image artifact removal method proposed by the present invention and the reconstructed signal are imaged by using the same ultrasonic imaging algorithm. Figure 3 (c) shows the ultrasonic pseudo-color B-scan image with artifacts, Figure 3 (d) shows the ultrasonic pseudo-color B-scan image without artifacts. Taking the image with artifacts as the standard, the peak signal-to-noise ratio and structural similarity of the image with removed artifacts and the standard image are calculated, which are 24.306 dB and 0.931 respectively. The image after artifact removal is not distorted. It can be seen that the ultrasonic image artifact removal method proposed by the present invention can effectively remove artifacts from the ultrasonic images of the standard defect test blocks of through-hole type, especially the near-surface artifact removal effect is obvious. Specific Embodiment 2

[0127] A standard defect test block of groove type is selected for the experiment. The test block and the geometric distribution of the defects are as Figure 4 (a) shown. The size of the straight groove is set to 5×3×2 mm 3 , the test block is made of aluminum material, and its ultrasonic sound speed is 6300 m / s. The ultrasonic sensor, detection method, imaging method, etc. used are the same as those in Embodiment 1. When using the adaptive ultrasonic image artifact removal method based on SoS function modulation proposed by the present invention, the time interval is 0.067 μs, and the number of echo signals of the signal K = 2. The original signal and the reconstructed signal are as Figure 4 (b) shown. Figure 4 (c) shows the ultrasonic pseudo-color B-scan image without removing artifacts, Figure 4(d) shows an artifact-removed ultrasonic pseudo-color B-scan image. The peak signal-to-noise ratio and structural similarity between the artifact-removed and standard images are 23.213 dB and 0.932, respectively. The image remains undistorted after artifact removal. This demonstrates that the proposed ultrasonic image artifact removal method can effectively remove artifacts from ultrasonic images of test blocks with standard groove defects.

[0128] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. For those skilled in the art, equivalent implementations and improvements that do not depart from the principles and technical spirit of the present invention are also within the scope of protection of the present invention.

Claims

1. An adaptive ultrasonic image artifact removal method based on SoS function modulation, characterized in that It includes the following steps: 1) Use an ultrasonic sensor to detect the test piece, receive the echo signal, and perform amplification, filtering, and envelope extraction on the echo signal to obtain the ultrasonic echo signal x(t) obtained from each scan; 2) Use the SoS function to perform amplitude modulation on the ultrasonic echo signal x(t) to obtain the transformed-domain signal y(t); In step 2), the SoS function is implemented by the weighted sum of the sinc function, and its frequency-domain expression is: where τ is the signal duration length; ω is the signal frequency; Ψ represents a set of consecutive integers, and the set length is determined by the information degrees of freedom of the signal; c m denotes the weighting coefficient, satisfying c m ≠ 0; Convert the SoS function to the time domain, and its expression is, The waveform of s(t) is determined by {c m} m∈Ψ and different modulation effects can be achieved with different parameter values; In step 2), use the SoS function to perform amplitude modulation on the ultrasonic echo signal x(t) to obtain the modulated transformed-domain signal y(t), where τ is the signal duration length; 3) Uniformly sample the transformed-domain signal y(t) to obtain the discrete sparse sampling sequence y[n] of the transformed-domain signal; In step 3), uniformly sample the transformed-domain signal y(t). The sampling frequency is determined according to the innovation rate of the signal. The sampling frequency is not less than the innovation rate of the signal. After sampling, the discrete sparse sampling sequence y[n] is obtained; The innovation rate is obtained in the following manner: Assume that the ultrasonic echo signal x(t) satisfies the condition that it can be characterized by a finite number of information degrees of freedom, then it can be expressed in a manner based on the information degrees of freedom, and its expression is, where c n,r and t n represent the amplitude parameter and time delay parameter of the signal respectively, is the basis function, and the continuous function C x (t1, t2) is used to represent the information degree of freedom of x(t) within the time interval [t1, t2]. The innovation rate is calculated based on the local time domain interval of the signal, that is, the finite information degree of freedom, and the signal can be truncated through a time window to obtain the local innovation rate P τ (t). The local innovation rate characterizes the number of information degrees of freedom of the signal within the duration interval τ; 4) Perform a discrete Fourier transform on the discrete sparse sampling sequence y[n], and its Fourier coefficients contain the information of the ultrasonic echo signal; 5) Apply a spectral estimation algorithm to estimate the parameters of the Fourier coefficients to obtain the time delay and amplitude parameters of the ultrasonic echo signal, and reconstruct the ultrasonic echo signal waveform in combination with the transmitted waveform of the ultrasonic sensor; 6) Use the reconstructed ultrasonic echo signal for acoustic imaging to obtain an artifact-free ultrasonic pseudo-color image; In step 4), express x(t) in the form of a Dirac flow signal u(t) based on the expression of the finite information degrees of freedom, which is expressed as, wherein, t k ∈ [0, τ), a k ∈ R, t k and a k are the echo signal delay and amplitude in the Gaussian model of the ultrasonic echo signal x(t); Expand the signal u(t) in a Fourier series, and its Fourier coefficients U[m] can be expressed in the form of a weighted sum of power series, that is, The Fourier coefficient U[m] contains the time delay and amplitude information of the ultrasonic echo signal x(t), that is, the Fourier coefficient information of the original ultrasonic echo signal; Step 5) specifically includes the following steps: 5.1) Solve the weighted value and exponent of the power series through a spectral estimation algorithm; 5.2) Convolve the filter coefficient h m with C[m], where C[m] = τU[m]. If the Z-transform of the nulling filter satisfies H(q k ) = 0, then the result of their convolution is zero, and the filter coefficient h m that meets the requirements can be obtained, expressed as: Let h0 = 1, and expand it to get: h1C[m - 1]+h2C[m - 2]+…+h K C[m - K]=-C[m] In the above formula, at least 2K consecutive Fourier series coefficients are required for the system of equations to have a unique solution; 5.3) Substitute the Fourier coefficients and the number of signal pulses into the system of equations in 5.2) to obtain the information of the parameters to be estimated Assume a continuous integer interval of length L, and Λ represents the value interval of the signal Fourier series coefficients, where Λ = [-l,..., 0,..., l], l is the number of pulses within τ, and L = 2l + 1. Then, the relationship between the number of Fourier coefficients L and the number of signal pulses K needs to satisfy: L = 2l + 1 ≥ 2K Obtain 2K consecutive Fourier coefficients and estimate the amplitude and time-delay parameters of the signal 5.4) According to the estimated parameters Combined with the Gaussian pulse model r(t) of the ultrasonic echo signal, adaptively reconstruct the ultrasonic echo signal that retains the important information of x(t); In step 5.1), the spectral estimation algorithm uses the annihilating filter method to solve the algorithm, including the following steps: 5.1.1) Construct an annihilating filter with a structural coefficient of Let where m ∈ Z, and the Z-transform of the annihilating filter is Let the zeros of H(z) be Let h0 = 1. After factorization, H(z) can be expressed as When all the delay parameters are different from each other, the zeros of the filter can uniquely represent the impulse delay parameters of the signal; 5.1.2) Obtain the coefficients h of the annihilating filter m , by finding the roots q of H(z) k , and obtain the delay parameter t k ; 5.1.3) Using U[m] to calculate the amplitude parameter a k Solve it.

2. The adaptive ultrasonic image artifact removal method based on SoS function modulation according to claim 1, characterized in that In step 1), the ultrasonic signal is generated by the high-voltage pulse excitation of the piezoelectric chip in the ultrasonic transducer. Assume that the Gaussian model of the ultrasonic echo signal x(t) is represented by r(t). The mathematical model of r(t) is: Among them, E represents the amplitude of the excitation pulse; η is its pulse width factor; f0 is the center frequency of the wafer; K represents the number of echoes; φ k is the initial phase; the parameters t k and a k carry the defect echo information of the test block to be detected, and the arrival time t k reflects the position information of the reflecting surface, and the echo amplitude a k reflects the sound intensity loss of the ultrasonic wave.

3. The adaptive ultrasonic image artifact removal method based on SoS function modulation according to claim 1, wherein, The reconstructed signal in step 6) is used to perform sound field calculation in the measured area, synthesize the sound field intensities of the spatial points, and represent the magnitude of the synthesized sound field intensity in color scale, thereby obtaining an ultrasonic pseudo-color image with artifacts removed.

Images