Method for realizing sound wave data detection by using laser speckle technology

Abstract

The invention relates to the technical field of acoustic signal detection and image processing, and discloses a method for realizing sound wave data detection by using a laser speckle technology, which comprises the following steps of: irradiating the surface of a detected target by using a laser to generate a speckle pattern; acquiring a speckle image sequence formed after laser irradiation through a high-speed image acquisition device; performing image processing on the speckle image sequence to estimate an optical flow velocity field of each image pixel point; estimating a target surface vibration frequency based on the optical flow velocity field; and estimating sound wave power based on the optical flow speed change amplitude and outputting corresponding frequency and power data. According to the invention, optical flow and frequency domain fusion analysis based on speckle images is adopted, so that synchronous detection of non-contact sound wave frequency and power is realized; a structure guide item is introduced, so that the optical flow estimation precision of a weak texture region is improved; and an optical flow module value energy model is constructed, and quantitative output of power is realized through calibration.

Description

Technical Field

[0001] The present invention relates to the technical field of acoustic signal detection and image processing, and in particular to a method for realizing acoustic wave data detection by utilizing laser speckle technology. Background Art

[0002] In practical engineering applications, the acquisition and analysis of acoustic wave propagation characteristics still primarily relies on contact measurement methods, such as capacitor microphones, piezoelectric accelerometers, and laser vibrometers. While these methods can obtain acoustic parameters such as frequency and amplitude, they generally suffer from complex deployment and high requirements for the target being measured. Contact sensors struggle to maintain stable operation, particularly in high-temperature, enclosed, remote, or vibration-sensitive environments, significantly reducing their reliability.

[0003] Currently, some non-contact solutions attempt to observe acoustic field disturbances using laser Doppler or high-speed photography. However, these approaches are limited to point measurements or rely on expensive specialized equipment. These approaches have high barriers to entry and limited system flexibility. Furthermore, most methods only extract frequency information and lack comprehensive modeling for energy intensity estimation, making them unable to fully describe the physical intensity of the acoustic field.

[0004] Among existing image-based methods, some studies have used optical flow to process speckle image motion, but these methods often employ traditional dense optical flow estimation models. This is prone to error propagation in image regions with weak texture and strong background noise. This approach fails to incorporate image structural information to aid modeling, resulting in a high probability of failure and insufficient stability in real-world speckle scenarios.

[0005] In terms of output, existing technologies tend to focus on single output modes, such as frequency spectrograms or image animations, and lack a structured interface for directly accessing results. This limits the in-depth application of non-contact detection systems in areas such as industrial platforms, intelligent control, and edge sensing, and makes it difficult to fully unlock the value of data. Summary of the Invention

[0006] In response to the shortcomings of the existing technology, the present invention provides a method for detecting acoustic wave data using laser speckle technology, which solves the problem that it is difficult to simultaneously achieve high-precision, non-contact acquisition of acoustic wave frequency and power in complex environments.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for detecting acoustic wave data using laser speckle technology, comprising the following steps:

[0008] S1. Using a laser to illuminate the surface of the target to generate a speckle pattern;

[0009] S2. Acquire a speckle image sequence formed after laser irradiation by a high-speed image acquisition device;

[0010] S3. performing image processing on the speckle image sequence to estimate the optical flow velocity field of each image pixel;

[0011] S4. Estimating the target surface vibration frequency based on the optical flow velocity field;

[0012] S5. Estimate the acoustic wave power based on the amplitude of the optical flow velocity change and output the corresponding frequency and power data.

[0013] Preferably, the wavelength range of the laser is 400 nanometers to 800 nanometers, the target surface is a solid material with microscopic roughness, and the speckle pattern is an intensity distribution image formed by the scattering interference of coherent laser light on the surface.

[0014] Preferably, the frame rate of the image acquisition device is not less than twice the required detection frequency range, the image acquisition time is not less than five times the maximum period of the acoustic wave signal, and the acquisition area is selected as the central symmetrical area of ​​the laser irradiation area.

[0015] Preferably, estimating the optical flow velocity field for the image sequence comprises the following steps:

[0016] S31. Construct an optical flow constraint equation based on the brightness preservation assumption between image sequences;

[0017] S32. Introducing the spatial optical flow continuity constraint to establish a two-dimensional variational optimization problem;

[0018] S33. Use an iterative numerical solution method to obtain the optical flow velocity vector of each pixel point, including the horizontal and vertical components.

[0019] Preferably, the two-dimensional variational optimization problem includes an image structure guidance term for suppressing excessive smoothing of image edge regions. The structure guidance term is weighted according to pixel grayscale gradient changes to enhance the accuracy of optical flow estimation in high-texture regions.

[0020] Preferably, the estimating the acoustic wave power includes:

[0021] S51 extracts the time-varying sequence of optical flow modulus values ​​for each pixel or image region;

[0022] S52. Performing frequency domain transformation on the time series;

[0023] S53. Extract the main peak frequency in the spectrum as the sound wave frequency.

[0024] Preferably, the acoustic wave power estimation is based on the square average of the optical flow modulus value per unit time, combined with the proportional coefficient obtained through experimental calibration, and outputs a dimensioned power value or normalized power distribution.

[0025] Preferably, the image processing process adopts a multi-scale image pyramid structure, performs optical flow estimation on images of different resolutions in sequence, and optimizes layer by layer from low to high to enhance the stability of optical flow estimation in multi-frequency interference and high noise scenes.

[0026] Preferably, the image area is divided into multiple sub-areas, and optical flow analysis and frequency extraction are performed on each sub-area respectively, and finally the frequency values ​​output by each sub-area are fused according to weights to generate an overall detection frequency result.

[0027] Preferably, the output of the sound wave data includes a frequency value, a power value under spatial coordinates, and a power distribution diagram in image form. The power distribution diagram is a two-dimensional spatial image, which is normalized to the maximum value for visual presentation.

[0028] The present invention provides a method for detecting acoustic wave data using laser speckle technology. It has the following beneficial effects:

[0029] 1. This invention employs an integrated detection strategy combining optical flow estimation based on speckle images with frequency-domain inversion, achieving a non-contact approach to acquiring acoustic frequency and power. Compared to existing approaches that rely on microphone arrays or structural sensors, this effectively avoids issues such as sensor interference and installation limitations associated with contact measurement, and addresses the technical barrier of effectively acquiring acoustic parameters in high-temperature, enclosed, and vibration-sensitive environments.

[0030] 2. This invention improves the accuracy of optical flow estimation in areas with complex and weak textures by constructing a variational optical flow model and introducing a structural guidance term. In existing methods, the inherent irregularities of speckle images often make accurate optical flow reconstruction difficult. However, this solution's guided smoothing strategy for image gradient responses addresses the error accumulation problem caused by structural uncertainty, thereby ensuring the prerequisites for frequency and power estimation throughout the system.

[0031] 3. This invention establishes a quantitative relationship between physical power and image velocity through experimental calibration. Unlike existing speckle analysis techniques, which focus solely on frequency identification, this approach further achieves quantitative output of acoustic wave power estimation, filling a technological gap in energy parameter identification for non-contact visual vibrometers.

[0032] 4. This invention utilizes a multi-format output mechanism, including frequency matrix, power image, and structured parameter package, making detection results compatible with a variety of industrial applications and analysis scenarios. Traditional systems often have a single output format, making them difficult to integrate directly into back-end platforms or control systems. This technical solution provides a highly open and adaptable result presentation method, resolving the issue of measurement result integration and utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a block diagram of the overall structure of the system of the present invention;

[0034] Figure 2 This is a flow chart of the acoustic wave data processing of the present invention;

[0035] Figure 3 Schematic diagram of the optical flow estimation and structure guidance mechanism of the present invention;

[0036] Figure 4 Extraction of optical flow modulus time series spectrum of the present invention;

[0037] Figure 5 Generate logic diagrams for the power estimation and thermal map of the present invention. DETAILED DESCRIPTION

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the specification of the present invention. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0039] Example:

[0040] Please see the attached Figure 1 -Attached Figure 5 Embodiments of the present invention provide a method for detecting acoustic wave data using laser speckle technology. This method can extract the frequency and power parameters of acoustic waves and is suitable for scenarios with mixed multi-frequency sound sources. The core concept is to generate surface speckle through laser irradiation and then capture the dynamic changes in speckle caused by acoustic wave disturbances through high-speed image acquisition. Furthermore, optical flow estimation is used to extract the speckle motion vector, which is then combined with frequency domain transformation to recover the acoustic wave frequency information. Finally, the acoustic wave power is calculated based on the energy of the optical flow changes.

[0041] This method requires the following hardware support:

[0042] A laser with an emission wavelength of 532nm (green light) or 650nm (red light);

[0043] A high-speed camera with a frame rate of at least 20,000 fps;

[0044] Frame grabber and processing computer;

[0045] The object being measured (metal plate, film material or hard plastic surface);

[0046] External sound source, such as piezoelectric speaker, ultrasonic actuator, etc.

[0047] Please see the attached Figure 1, the method comprises the following steps:

[0048] S1. Using a laser to illuminate the surface of the target to generate a speckle pattern;

[0049] The laser shines through a focusing lens onto the surface of the object being measured, creating a typical speckle pattern. The camera is positioned at a fixed angle to the laser to avoid strong reflections. The entire imaging system must be placed on a stable stand to prevent vibration interference.

[0050] The camera captures images at a high continuous frame rate. Image size is kept at or below 640×480 to improve frame rate and reduce computational overhead. Images are saved as uncompressed grayscale frame sequences to reduce subsequent errors.

[0051] S2. Acquire a speckle image sequence formed after laser irradiation by a high-speed image acquisition device;

[0052] Please see the attached Figure 2 The technical approach involved in this invention focuses on the microscopic vibrations generated on the surface of an object under test by acoustic waves. This microvibration information is captured through laser speckle imaging, and the acoustic frequency and power are then inverted using optical flow estimation and frequency domain analysis. Throughout the entire technical implementation process, image acquisition and preprocessing are at the forefront of the signal chain, and their quality directly impacts the accuracy and stability of subsequent optical flow analysis and spectrum extraction.

[0053] To ensure the effectiveness of optical images, the design of this module not only includes the rational configuration of the laser optical path and imaging angle, but also covers the control of key parameters such as image frame rate, sampling duration, and noise suppression. It also pre-processes the raw image data using methods such as grayscale enhancement and time-domain differencing. The following detailed description of the technical aspects of image acquisition and pre-processing is combined with implementation examples.

[0054] In this embodiment, the image acquisition process includes two sub-steps: laser irradiation setting and high-speed image acquisition.

[0055] In general, a laser should have a stable output wavelength, with continuous lasers typically using wavelengths of 532 nanometers (for green lasers) or 650 nanometers (for red lasers). The surface being measured must have a certain degree of microscopic roughness to produce a strong interference speckle pattern. The laser beam is focused onto the target surface through a collimating optical system, forming a stable speckle pattern.

[0056] The angle between the camera's imaging optical axis and the laser's incident direction can be set within a range of 15° to 45°, which helps form a strong speckle interferometry image while reducing interference from reflected light on the imaging system. In one specific embodiment, the image acquisition device uses a high-speed CMOS camera with a frame rate of 20,000 frames per second and an image resolution of 512×512 pixels.

[0057] In actual operation, in order to ensure the accuracy of frequency extraction, the camera sampling frame rate f s Satisfy the Nyquist sampling criterion, that is:

[0058] f s ≥2f max ;

[0059] in:

[0060] f s Indicates the acquisition frame rate in Hertz (Hz);

[0061] f max Indicates the maximum frequency component of the measured sound wave signal, in Hertz (Hz).

[0062] In some embodiments, in order to obtain better spectral resolution, the total sampling time T is set to be more than 5 times the period of the measured sound wave, and the number of sampling frames N can be calculated as:

[0063] N=T·f s ;

[0064] in:

[0065] N is the total number of frames (unit: frame);

[0066] T is the total sampling time, in seconds (s);

[0067] f s is the sampling frame rate in Hertz (Hz).

[0068] After the image data acquisition is completed, the system obtains the image sequence I(x, y, t n ),in:

[0069] I is the grayscale value, the unit is dimensionless (usually 8-bit or 12-bit integer);

[0070] x,y are spatial pixel coordinates, in pixels (px);

[0071] t n =n / f s , is the timestamp of the n-th frame image, in seconds (s).

[0072] In this embodiment, image preprocessing includes three processes: noise filtering, temporal difference enhancement, and image normalization.

[0073] Because speckle images contain high-frequency noise, direct optical flow estimation typically results in computational instability. To reduce the impact of local grayscale fluctuations on analysis, a median filter is used to spatially smooth each image frame. The median filter window size, w×w, can be set to 3×3 or 5×5 pixels and is dynamically adjusted based on the image texture complexity.

[0074] As an option, a Gaussian filter can also be used for spatial noise reduction, with the kernel function standard deviation σ g The value of is generally set to 0.8 to 1.2.

[0075] In a possible implementation, in order to enhance the dynamic characteristics of the image over time, the inter-frame difference image D(x, y, t n ), defined as:

[0076] D(x,y,t n )=|I(x,y,t n )-I(x,y,t n-1 )|;

[0077] This difference operation is used to highlight the brightness changes of the speckle caused by the vibration of the sound wave. After this operation, the grayscale image is still maintained and serves as an auxiliary input for the subsequent optical flow estimation.

[0078] Specifically, in order to meet the numerical requirements of the image processing module, all image grayscale values ​​are normalized and the grayscale range is adjusted to the interval [0, 1] using the following formula:

[0079]

[0080] in: Represents the normalized grayscale value; I min with I max They represent the minimum and maximum grayscale values ​​of the image sequence respectively; I is the original grayscale value, and the unit is the same as above.

[0081] In some implementations, grayscale normalization not only improves the stability of subsequent calculations, but also helps avoid the algorithm's sensitivity to changes in the image brightness range.

[0082] Furthermore, to improve processing accuracy and efficiency, this embodiment may also introduce a region of interest (ROI) selection mechanism.

[0083] This mechanism allows for the selection of a region within an image for analysis based on the direction of acoustic radiation, the laser illumination area, or the brightness response of a difference image. Image filtering, normalization, and optical flow modeling are then performed within the selected region. This approach reduces computational effort while enhancing the algorithm's adaptability to multi-region scenarios.

[0084] As a possible processing strategy, a background reference frame can be constructed for the image sequence, and background modeling can be performed between each frame and the reference frame to further suppress non-periodic interference and system drift and improve the signal-to-noise ratio.

[0085] In summary, the image acquisition and preprocessing module provided in this embodiment fully combines optical imaging, image processing, and frequency domain sampling theory, and is stable and versatile. The technical details disclosed above satisfy the requirements of those skilled in the art to implement this technical solution based on the contents of the specification.

[0086] If you need to select a higher frame rate based on the actual scene, adapt to different material surfaces, or add depth filter processing, this module also has good scalability and adaptability.

[0087] S3. performing image processing on the speckle image sequence to estimate the optical flow velocity field of each image pixel;

[0088] Please see the attached Figure 3 -Attached Figure 4 After image preprocessing, the system obtains a grayscale image sequence with stable structure and clear texture. The next step is to enter the optical flow field estimation process. This step is the core processing link of the present invention. Its purpose is to calculate the pixel displacement rate field formed by the time-varying speckle pattern in the time series image.

[0089] Grayscale changes in pixels within an image reflect the minute dynamic disturbances on the surface of the object being measured due to acoustic waves. Optical flow estimation can quantify the motion vector information during this disturbance. Subsequent frequency and power extraction, in turn, relies on the accuracy and continuity of the optical flow field. Therefore, optical flow estimation constitutes the fundamental data source in the acoustic wave detection path of this invention.

[0090] In this embodiment, the optical flow field estimation process establishes a two-term optimization model based on the variational method principle, and performs constraint modeling through the brightness preservation relationship between image sequences.

[0091] In general, it is assumed that the brightness value of the same target remains unchanged in two consecutive frames of images. Based on this assumption, the following brightness consistency constraint is constructed:

[0092] I(x+u x ,y+u y ,t+Δt)≈I(x,y,t);

[0093] Where: I(x,y,t) represents the grayscale value of the pixel at the time t (x,y); u x 、u y represent the pixel displacement along the x and y directions respectively; Δt is the time interval between frames, in seconds; the domains of all variables are continuous or differentiable function spaces.

[0094] In order to express the motion rate more clearly, the optical flow velocity vector is introduced:

[0095] v(x,y,t)=(v x (x,y,t),v y (x,y,t)) T ;

[0096] in: is the displacement per unit time in the x direction (i.e., velocity); is the velocity in the y direction; T represents the column vector transpose.

[0097] Expand the brightness consistency condition in the first order Taylor at Δt→0 to obtain the basic constraint equation of optical flow:

[0098]

[0099] in: Represents the image spatial gradient; is the time derivative of the image; represents the vector inner product.

[0100] In this embodiment, in order to obtain a complete optical flow velocity field, a variational modeling approach is introduced to construct an energy functional including a data consistency term and a regularized smoothing term.

[0101] Specifically, the objective function E(v) represents the cost function of the optical flow field over the entire image area, and is in the following form:

[0102]

[0103] Where: Ω is the image definition area, the unit is pixel area; Indicates v x The square of the gradient norm; The same definition; is the regularization coefficient, which is used to balance the relative weights of the smoothing term and the data term.

[0104] In one possible implementation, to further improve the accuracy of optical flow estimation in edge regions, an image structure guidance factor can be introduced into the regularization term. The structure guidance weight is constructed based on the image gradient intensity to avoid over-smoothing of edge regions. In this case, the energy function can be expanded to:

[0105]

[0106] in: is the structural weight factor; σ is the parameter that controls weight attenuation, and its unit is the grayscale change scale; exp(·) is the natural exponential function; when the local gradient of the image is large, w approaches 0, weakening the regularization strength; when the image is smooth, w→1, strengthening the smoothness constraint of the optical flow field.

[0107] In actual implementation, this embodiment uses iterative optimization to solve the above variational problem and introduces an image pyramid structure to enhance stability.

[0108] In general, due to the large actual motion or complex image details, directly solving the optical flow on the original image will result in mismatch. Therefore, in actual operation, a three-layer image pyramid is used to iteratively solve the optical flow field from bottom to top.

[0109] After scaling each layer of images, a rough optical flow estimate is generated, which is then upsampled to the next layer via bilinear interpolation as the initial value. This strategy effectively avoids falling into local minima and improves the stability of the overall solution.

[0110] Furthermore, in some embodiments, the modulus value after optical flow solution can be calculated as:

[0111]

[0112] Where: v mag is the optical flow modulus value, in pixels per second (px / s); v x 、v y are the transverse and longitudinal velocity components, defined as before;

[0113] The modulus value is the plane velocity of the pixel point per unit time.

[0114] This quantity will be used as the core input data in subsequent frequency and power estimation for timing analysis and energy integration.

[0115] Overall, this module is fully transparent, from physical modeling and mathematical expression to numerical solution. The formula structure and character symbols are unambiguous, meeting the requirements for technical replicability. The regularization parameter and number of image pyramid layers can be flexibly selected based on different application scenarios, further adapting to surface materials with different texture characteristics or vibration amplitude ranges.

[0116] S4. Estimating the target surface vibration frequency based on the optical flow velocity field;

[0117] After completing the calculation of the optical flow field, the system has obtained the motion velocity field of the measured surface at different time nodes. At this time, each pixel point or area corresponds to a vector change sequence with respect to time. Since the sound wave is a periodic excitation source, if it acts on the surface of the object, it will cause continuous, slight grayscale disturbances in the speckle pattern. These disturbances have been converted into time series information through the optical flow modulus value, and the corresponding frequency components can be extracted in the frequency domain. Frequency analysis is based on this foundation, and its main goal is to identify the dominant frequency of the sound wave and the possible multi-frequency structure as input parameters for subsequent power estimation and sound field inversion.

[0118] In this embodiment, the frequency analysis is based on a time series constructed by the change of the optical flow modulus value over time, and the main frequency component is extracted through discrete Fourier transform.

[0119] In general, the optical flow vector v(x,y,t)=(v x ,v y ) T Represents the velocity vector of the pixel (x, y) at time t. To perform frequency domain analysis, its modulus must be calculated first:

[0120]

[0121] Where: v mag (x,y,t): represents the optical flow modulus value of the pixel (x,y) at time t, in pixels per second (px / s); v x (x, y, t): horizontal and vertical optical flow components, respectively, with the same units as above; x, y: two-dimensional spatial coordinates in the image, in pixels; t: time variable, in seconds (s).

[0122] In one possible implementation, the optical flow modulus value at a fixed position (x0, y0) in the image can be selected to form a time series:

[0123] s(t n )=v mag (x0,y0,t n );

[0124] Where: t n =n / f s , represents the acquisition time of the nth frame image; f s : image sampling frame rate, in Hertz (Hz); n = 0, 1, ..., N-1, is the frame sequence number; N: total number of frames, satisfying N = T f s , where T is the sampling time.

[0125] In this embodiment, the fast Fourier transform (FFT) is used to transform the time series s(t n ) to perform frequency domain conversion.

[0126] The transformation result can be expressed as:

[0127]

[0128] Where: S(f k ):frequency f k The complex spectrum response at ; is the kth frequency sampling point;

[0129] i: imaginary unit; k = 0, 1, ..., N-1.

[0130] The spectrum intensity is defined as the spectrum modulus:

[0131]

[0132] In: Re(S(f k )):real part; Im(S(f k )): is the imaginary part.

[0133] Specifically, in the frequency extraction process, the spectrum intensity |S(f k )|’s main peak point, you can get the main frequency information.

[0134] In a typical implementation:

[0135]

[0136] Where: f0: is the detected main frequency, in Hertz (Hz); argmax: represents the frequency point that maximizes the spectrum intensity.

[0137] As an option, if there are multiple significant peaks (typically more than 60% of the maximum amplitude), multiple frequencies can be extracted:

[0138] {f1,f2,...,f M}={f k ∈[0,f s / 2]||S(f k )|≥τ·max|S(f k )|};

[0139] Where: τ: the amplitude threshold factor of multi-frequency recognition (recommended setting is 0.6 to 0.8); M: the number of frequency components identified; f s / 2: Nyquist frequency.

[0140] In some embodiments, to reduce spectrum leakage effects, a windowing function may be applied to the time series.

[0141] Common window functions include Hamming window, Hanning window, etc. Take the Hamming window as an example:

[0142]

[0143] After windowing, the time series is corrected to:

[0144] s ′ (t n )=s(t n )·w(n);

[0145] This operation can make a trade-off between the spectrum main lobe width and side lobe leakage, which is beneficial to improving the resolution and stability of main frequency detection.

[0146] In this embodiment, it can also be expanded to a regional spectrum averaging strategy.

[0147] That is, for the image area The pixel optical flow modulus value sequence within is uniformly spectrally summed and averaged:

[0148]

[0149] in: is the number of pixels in the area; S x,y (f k ): is the spectrum corresponding to the pixel point (x, y); is the regional average spectrum.

[0150] Final frequency extraction result:

[0151]

[0152] This strategy is suitable for regional analysis scenarios with high texture consistency and low signal-to-noise ratio, and helps improve the noise resistance of frequency analysis.

[0153] In summary, this module constructs a time-domain representation of the frequency signal based on the temporal variations of the optical flow modulus values. It extracts frequency components using a high-resolution spectrum transform algorithm. This module provides comprehensive primary frequency extraction and multi-frequency resolution capabilities while meeting Nyquist conditions and temporal resolution requirements. The mathematical definitions, parameter notations, and unit dimensions of the analysis process are publicly available, and the technical structure is clear, ensuring reproducibility and implementability.

[0154] S5. Estimate the acoustic wave power based on the amplitude of the optical flow velocity change and output the corresponding frequency and power data.

[0155] Please see the attached Figure 5 After completing frequency extraction, the system has clearly identified the dominant frequency or frequency component sequence of the sound wave. However, frequency information is not sufficient to fully characterize the intensity distribution of the sound field. In order to further describe the energy characteristics of the sound wave acting on the surface of the object, it is necessary to introduce a power estimation module to quantify the sound wave energy. Power estimation is based on the optical flow modulus value and uses the energy density model to estimate the equivalent power intensity generated by the sound wave action per unit time and per unit area, providing a complete description of the sound wave physical field for the present invention.

[0156] In this technical path, power estimation is not based on the traditional microphone sound pressure calculation model, but combines the square of the modulus and integral average of the optical flow velocity field to construct an optical speckle response model equivalent to the sound wave energy, which is suitable for estimating small-amplitude surface vibration fields in non-contact scenarios.

[0157] In this embodiment, the acoustic wave power estimation constructs an energy expression model based on the time average value of the square of the optical flow modulus value.

[0158] In general, the optical flow modulus v mag (x, y, t) represents the two-dimensional velocity amplitude of the pixel point (x, y) per unit time, which is defined by the above formula:

[0159]

[0160] Where: v mag (x,y,t): Optical flow modulus value, in pixels per second (px / s); v x (x,y,t),v y (x, y, t): The velocity components of the pixel in the horizontal and vertical directions, in pixels per second (px / s); x, y: spatial pixel position, in pixels (px); t: time variable, in seconds (s).

[0161] Based on the above definition, a unit time energy expression model is constructed. Specifically, the acoustic wave power estimation uses the following integral expression:

[0162]

[0163] Where: P(x,y): Equivalent power of acoustic wave per unit area at pixel point (x,y), in watts (W); κ: Power conversion factor, in W·s 2 / px 2 , used to convert the square of pixel velocity into real physical quantity; T: integration interval, that is, total sampling time, in seconds (s); v mag (x,y,t): Optical flow modulus per unit time, as described above.

[0164] In a possible implementation, the above integral can be discretized into a summation form, which is applicable to the discrete time structure of the frame sequence.

[0165] At the image frame rate f s Under the condition of, the total number of frames is N, and the sampling time is T = N / f s , then:

[0166]

[0167] Where: t n =n / f s : Indicates the time point of the nth frame; all other parameters are defined as before.

[0168] As an option, in order to avoid estimation deviation caused by brightness difference or texture structure in different areas, the whole image or a certain area can be The power results within are normalized:

[0169]

[0170] Where: P norm (x,y): normalized power value, unit is dimensionless; maxP(i,j): the maximum power value in the selected area;

[0171] The normalized results are convenient for image visualization output and are suitable for heat map or pseudo color coding processing.

[0172] In this embodiment, the power conversion coefficient κ is obtained through experimental calibration.

[0173] Specifically, a standard signal source can be selected, such as a piezoelectric ultrasonic transmitter with a fixed frequency and amplitude, and its excitation power can be set to P ref , and confirmed by external sound intensity measurement equipment. In the experiment, the optical flow modulus data corresponding to the excitation is collected and its average square value is calculated:

[0174]

[0175] Then solve for the calibration coefficients:

[0176]

[0177] This process is repeated under multiple stimulus conditions, and the average value is taken as the system parameter and embedded in the power estimation module. This calibration process only needs to be performed once and is applicable to all subsequent measurement tasks using the same device and parameter configuration.

[0178] In some embodiments, a power time variation sequence may be further constructed to enable monitoring and analysis of instantaneous energy variations.

[0179] The instantaneous power per frame is defined as:

[0180]

[0181] The power time series response is then constructed through moving average filtering to detect dynamic energy fluctuations, such as the energy envelope of high-frequency events such as impact sounds and short-term excitations.

[0182] In summary, the power estimation module converts pixel velocity into equivalent acoustic energy by combining physical modeling, optical flow field characteristics, and time series statistics. The technical approach is clear, the mathematical expression is explicit, and all parameters are fully disclosed. This module, along with image acquisition, optical flow estimation, and frequency extraction, forms a logically closed loop, meeting the engineering requirements for contactless measurement of multiple parameters in acoustic signals and exhibiting excellent scalability and adaptability.

[0183] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for detecting acoustic wave data using laser speckle technology, characterized in that: The following steps are involved: S1. Using a laser to illuminate the surface of the target to generate a speckle pattern; S2. Acquire a speckle image sequence formed after laser irradiation by a high-speed image acquisition device; S3. performing image processing on the speckle image sequence to estimate the optical flow velocity field of each image pixel; S4. Estimating the target surface vibration frequency based on the optical flow velocity field; S5. Estimate the acoustic wave power based on the amplitude of the optical flow velocity change and output the corresponding frequency and power data.

2. The method for realizing acoustic wave data detection using laser speckle technology according to claim 1, characterized in that: The wavelength range of the laser is 400 nanometers to 800 nanometers, the target surface is a solid material with microscopic roughness, and the speckle pattern is an intensity distribution image formed by the scattering interference of coherent laser light on the surface.

3. The method for realizing acoustic wave data detection using laser speckle technology according to claim 1, characterized in that: The frame rate of the image acquisition device is not less than twice the required detection frequency range, the image acquisition time is not less than five times the maximum period of the acoustic wave signal, and the acquisition area is selected as the central symmetrical area of ​​the laser irradiation area.

4. The method for realizing acoustic wave data detection using laser speckle technology according to claim 1, characterized in that: Estimating the optical flow velocity field for the image sequence comprises the following steps: S31. Construct an optical flow constraint equation based on the brightness preservation assumption between image sequences; S32. Introducing the spatial optical flow continuity constraint to establish a two-dimensional variational optimization problem; S33. Use an iterative numerical solution method to obtain the optical flow velocity vector of each pixel point, including the horizontal and vertical components.

5. The method for realizing acoustic wave data detection using laser speckle technology according to claim 4, characterized in that: The two-dimensional variational optimization problem includes an image structure guidance term for suppressing excessive smoothing in the edge areas of the image. The structure guidance term is weighted according to the change in pixel grayscale gradient to enhance the accuracy of optical flow estimation in high-texture areas.

6. The method for realizing acoustic wave data detection using laser speckle technology according to claim 1, characterized in that: The estimated sound wave power includes: S51 extracts the time-varying sequence of optical flow modulus values ​​for each pixel or image region; S52. Performing frequency domain transformation on the time series; S53. Extract the main peak frequency in the spectrum as the sound wave frequency.

7. The method for realizing acoustic wave data detection using laser speckle technology according to claim 1, characterized in that: The acoustic wave power estimation is based on the square average of the optical flow modulus value per unit time, combined with the proportional coefficient obtained through experimental calibration, and outputs a dimensioned power value or normalized power distribution.

8. The method for realizing acoustic wave data detection using laser speckle technology according to claim 1, characterized in that: The image processing process adopts a multi-scale image pyramid structure to perform optical flow estimation on images of different resolutions in sequence, and optimizes layer by layer from low to high to enhance the stability of optical flow estimation in multi-frequency interference and high noise scenes.

9. The method for realizing acoustic wave data detection using laser speckle technology according to claim 1, characterized in that: The image area is divided into multiple sub-areas, and optical flow analysis and frequency extraction are performed on each sub-area. Finally, the frequency values ​​output by each sub-area are fused according to the weights to generate the overall detection frequency result.

10. The method for realizing acoustic wave data detection using laser speckle technology according to claim 1, characterized in that: The output of the acoustic wave data includes a frequency value, a power value in spatial coordinates, and a power distribution diagram in image form. The power distribution diagram is a two-dimensional spatial image that is normalized to its maximum value for visualization.

Images