Ultrasonic microscopic image restoration method based on high-frequency focusing probe sound field calibration
Through calibration and denoising of sound pressure distribution, the blur problem caused by sound beam overlap and focus mismatch of ultrasonic microscope images is solved, which improves the contrast and resolution of imaging, ensuring the clarity of feature edges and the accuracy of shapes. Identify.
Patent Information
- Application Number
- CN202411131940.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2024-11-26
AI Technical Summary
During the movement of the ultrasonic microscopy image, the edges of the feature are blurred due to the overlap of the sound beams, making it difficult to identify the precise position and shape. When the focus area does not match the measured target scale, the imaging results are blurred, making it difficult to identify the precise position and shape of the feature.
By acquiring ultrasonic microscope scanning images and measurement data, post-processing includes normalization, grayscale and denoising, using hydrophone frequency response deconvolution to calculate the sound pressure distribution, adjust the distance between the probe and the measured target, and use The bilateral filter denoising and convolutionally treats the sound pressure distribution as a fuzzy kernel to restore the ideal image.
Effectively eliminate the degradation of imaging quality caused by the sound pressure distribution of ultrasonic microscope probes, improve the contrast and resolution of ultrasonic microscope scanning imaging, and ensure the clarity of feature edges and accurate recognition of shapes.
Smart Images

Figure CN119027350A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of image processing, and in particular to an ultrasonic microscopic image restoration method based on high-frequency focusing probe sound field calibration. Background Art
[0002] As an advanced imaging technology, ultrasonic microscopy is widely used in many fields. Different from traditional optical imaging technology, this technology is unique in that it can provide internal structural information at a microscopic scale. Its principle is to use ultrasonic microprobes to generate sound waves to penetrate materials. When passing through the interface between two different materials, due to the different acoustic impedances of different materials, the absorption and reflection of sound waves are different. The reflected / penetrated ultrasonic energy information or phase information changes are collected to check for defects such as delamination, cracks or voids inside the material. Therefore, the ultrasonic scanning image also reflects the distribution of acoustic impedance changes on the measured plane.
[0003] During the imaging process, when the focused ultrasound microprobe scans along the moving direction, the focal plane sound pressure distribution will scan the characteristic plane as the probe moves. As the projection area of the focal plane sound pressure distribution on the characteristic surface increases, the echo amplitude received by the probe will also increase. When the probe performs stepping motion, each step measures and records the energy of the reflected signal, which is then quantified and mapped into pixel values. When the step distance is less than the beam width, adjacent steps will cause the two beams to overlap in a certain area, resulting in the overlapping area being mapped as components of different pixel values in the image. This repeated mapping allows the material features to be expanded in the pixel mapping. The striped column represents the grayscale value mapped to the reflected signal, and its brightness and height represent the size of the grayscale value. The pixel value distribution does not fully reflect the shape of the feature, but changes slowly near the edge of the feature, making the edge and details of the feature blurred between multiple pixels, making the precise position and shape of the feature difficult to identify. The dense paving condition is met during ultrasonic scanning, that is, the beams overlap to prevent missed detection.
[0004] When the focusing area of the focusing probe is not fine enough compared to the spatial scale of the measured target, the imaging result becomes blurred due to the convolution effect of the spatial distribution of sound pressure and the characteristic acoustic impedance of the measured target, resulting in a decrease in the contrast of the precise position and shape of the feature and making it difficult to identify. Therefore, a method for ultrasonic microscopic image restoration based on high-frequency focusing probe sound field calibration is needed. Summary of the invention
[0005] The object of the present invention is to provide an ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration.
[0006] To achieve the above object, the present invention is implemented according to the following technical solutions:
[0007] The first aspect of the present invention provides an ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration, comprising the following steps:
[0008] The following steps are involved:
[0009] Acquire a scanning image of an ultrasonic microscope and measurement data of a microprobe, wherein the measurement data includes a center frequency, a focal length, and an effective radius;
[0010] Post-processing the ultrasonic microscope scanning image, wherein the post-processing includes normalization and grayscale processing;
[0011] The center frequency of the microprobe is obtained by deconvolution of the hydrophone frequency response, the focal length is calculated according to the transit time and sound velocity of the maximum axial sound pressure, the effective radius is obtained according to the axial sound pressure distribution measurement, and the sound pressure distribution G(g,h) is determined by the measurement data;
[0012] The image is denoised, and the sound pressure distribution G(g,h) is used as a blur kernel for convolution. The difference between the convolution result and the denoised image is taken, and the size of the error image is calculated to determine whether the error size is lower than the threshold. The error determination formula is as follows:
[0013]
[0014] If the error is lower than the threshold, it means that the difference between the denoised image and the ideal image is small, and the two can be considered equivalent;
[0015] If the error is higher than the threshold, the update step is entered, and the update formula is:
[0016]
[0017] r n represents the nth restored image, G * represents the flipping of the matrix G about the origin, C is a small constant added to prevent the denominator from being zero, and the error of the above steps is iterated to be lower than the threshold b n , then the ideal image is output.
[0018] Furthermore, the method for obtaining a scanning image of an ultrasonic microscope includes:
[0019] Adjust the distance between the probe of the ultrasonic microscope and the object to be measured so that the focusing plane of the sound beam of the probe coincides with the plane of the characteristic to be measured;
[0020] Manually set the time range of the characteristic signal in the reflected echo as the gate threshold;
[0021] Calculate the positive peak value / negative peak value / peak-to-peak value of the signal in the threshold as the grayscale value of the current position;
[0022] The probe of the mobile ultrasonic microscope is scanned across the entire surface, and the above process is repeated to image the entire plane.
[0023] Furthermore, the method for measuring the effective radius according to the axial sound pressure distribution includes adjusting the relative position of the needle / thin membrane hydrophone and the ultrasonic microprobe to ensure that the two are located on the same horizontal axis, and scanning along the axis to at least include the axial sound pressure maximum point to obtain the axial sound pressure distribution;.
[0024] Furthermore, the synchronization signal output of the pulse transmitter and receiver is connected to an oscilloscope, and the time difference between the synchronization signal and the pulse signal received by the hydrophone is recorded, and the focal length is calculated through the time difference; the recorded time difference is multiplied by the speed of sound in water at the test temperature to obtain the focal length of the probe, and the spectrum analysis is performed on the signals collected by the electric pulse and the hydrophone, and the center frequency of the probe is calculated after deconvolution processing using the frequency response data of the hydrophone.
[0025] Furthermore, the method for denoising the image includes linearly mapping the pixel values collected by scanning to a range of 0 to 1, using a bilateral filter to smooth the image noise while retaining important edge information, and using f s Spatial domain function and f r The range function controls the balance between smoothing noise and preserving edges. The specific formula for the denoised image is:
[0026]
[0027] where Ω is a neighborhood centered at s, ||ss′|| is defined as the Euclidean distance between pixels s and s′, and |I(s)-I(s′)| represents the intensity difference between pixels s and s′;
[0028] The spatial domain is
[0029]
[0030] The range domain is:
[0031]
[0032] Adjust the neighborhood Ω of the spatial domain and the standard deviation σ of the range domain r There are two parameters, the size of the neighborhood controls the degree of spatial smoothing, and the neighborhood is adjusted in the range of [2-5] pixels in steps of 1. A larger neighborhood will result in more spatial smoothing, but may blur details.
[0033] The standard deviation controls the smoothness of the grayscale value intensity. The standard deviation is adjusted in the range of [0-2] with a step of 0.1. A larger standard deviation makes the filter insensitive to brightness differences and loses brightness details, but can better remove noise. This makes it possible to effectively remove additive noise from the original data without blurring texture details.
[0034] Furthermore, when the sound pressure distribution G(g,h) is used as the blur kernel for convolution processing, it is described by the convolution of the sound pressure distribution function of the working plane of the focusing probe and the distribution of the acoustic impedance change of the measured plane:
[0035]
[0036] Among them, G(g,h) is the sound pressure distribution function, X(g,h) represents the distribution of acoustic impedance changes in the measured plane, N(g,h) represents the additive noise introduced by the system, and g and h represent the physical quantities of the actual distance.
[0037] In order to obtain an ideal unblurred two-dimensional image, it is necessary to deconvolve the equation, and the modeling result of G(g,h) is:
[0038]
[0039] Among them, f is the center frequency of the sound wave, a is the effective radius of the probe, c represents the speed of sound in the medium, l f Indicates the focal length of the probe.
[0040] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:
[0041] The present invention calibrates the acoustic pressure distribution of the focal plane of the ultrasonic microprobe and applies an image restoration method to eliminate the degradation of imaging quality caused by the acoustic pressure distribution of the ultrasonic microprobe, thereby improving the contrast and resolution of ultrasonic microscope scanning imaging. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 A flowchart of the steps of an ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration according to the present invention;
[0043] Figure 2 The present invention provides an ultrasonic microscopy working mode of an ultrasonic microscopy image restoration method based on high-frequency focused probe acoustic field calibration;
[0044] Figure 3 A schematic diagram of a method for measuring the acoustic field characteristics of an ultrasonic microprobe in an ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration of the present invention;
[0045] Figure 4 The present invention is a schematic diagram of an image restoration strategy of an ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration. DETAILED DESCRIPTION
[0046] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0047] Reference Figure 1 As shown, the first aspect of the present invention provides an ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration, comprising:
[0048] The purpose of the present invention is mainly achieved through the following technical solutions:
[0049] Step 1: Obtain a scanning image using an ultrasonic microscope;
[0050] Step 2: Measure the parameters of the ultrasonic microprobe used in the scanning process, obtain the center frequency f of the microprobe based on the hydrophone frequency response deconvolution, and calculate the focal length l based on the transit time and sound velocity of the maximum axial sound pressure. f , the effective radius a is obtained based on the axial sound pressure distribution measurement;
[0051] Step 3: Obtain the sound velocity c based on the medium and temperature used, and determine the sound pressure distribution G(g,h);
[0052] Step 4: normalize the grayscale value of the ultrasonic microscope scanning image;
[0053] Step 5: Use a bilateral filter to denoise the grayscale image to remove the additive noise N(g,h) introduced by the system;
[0054] Step 6: Use G(g,h) constructed in step 2 as the blur kernel and use the Lucy-Richardson method to deblur the image to obtain the restored image.
[0055] In step 1, firstly, the distance between the probe and the target needs to be adjusted so that the focusing plane of the probe's acoustic beam coincides with the plane of the feature to be measured, and the time range of the feature signal in the reflected echo is manually set as the gate threshold; the positive peak value / negative peak value / peak-to-peak value of the signal in the threshold is calculated as the gray value of the current position;
[0056] The probe of the mobile ultrasonic microscope is scanned across the entire surface, and the above process is repeated to image the entire plane. At the beginning of the scan, it is necessary to set a reasonable scanning speed and scanning step according to the actual working conditions.
[0057] Reference Figure 2and 3 As shown, in step 2, a needle / film hydrophone is used to measure the center frequency f and focal length l of the ultrasonic microprobe. f and effective radius a. Use a pulse transmitter and receiver to connect the ultrasonic microprobe and fix the ultrasonic microprobe in the water. Set appropriate energy and bandwidth parameters to excite the probe. Carefully adjust the relative position of the needle / thin membrane hydrophone and the ultrasonic microprobe, including the distance and alignment between the two. Ensure that the receiving sensitive element of the hydrophone is in the center of the focal area of the ultrasonic microprobe sound field by finding the maximum axial sound pressure. Connect the synchronization signal output by the pulse transmitter and receiver to the oscilloscope, record the time difference between the synchronization signal and the pulse signal received by the hydrophone, and then multiply it by the speed of sound in water at that temperature to obtain the focal length l. f ; According to the spectrum of the electric pulse and the hydrophone signal, the center frequency f is calculated by deconvolution of the hydrophone frequency response;
[0058] In step 3, according to the measured focal length l f and center frequency f. On the basis of correcting the spatial averaging effect introduced by the effective size of the hydrophone, the effective radius a of the probe is calculated using the axial sound pressure distribution, and the sound pressure distribution G(g,h) can be determined in combination with step 2.
[0059] In step 4, the pixel values collected by scanning are linearly mapped to the range of 0 to 1. The bilateral filter smoothes the image noise while retaining important edge information. s Spatial domain function and f r The range function controls the balance between smoothing noise and preserving edges. The specific formula for the denoised image is:
[0060]
[0061] where Ω is a neighborhood centered at s, ||ss′|| is defined as the Euclidean distance between pixels s and s′, and |I(s)-I(s′)| represents the intensity difference between pixels s and s′.
[0062] The spatial domain is
[0063]
[0064] The range domain is:
[0065]
[0066] Adjust the neighborhood Ω of the spatial domain and the standard deviation σ of the range domain r There are two parameters, the size of the neighborhood controls the degree of spatial smoothing, and the neighborhood is adjusted in the range of [2-5] pixels in steps of 1. A larger neighborhood will result in more spatial smoothing, but may blur details.
[0067] The standard deviation controls the smoothness of the grayscale value intensity. The standard deviation is adjusted in the range of [0-2] with a step of 0.1. A larger standard deviation makes the filter insensitive to brightness differences and loses brightness details, but can better remove noise. This makes it possible to effectively remove additive noise from the original data without blurring texture details.
[0068] Reference Figure 4 As shown, in step 6, the denoised image is convolved with the sound pressure distribution G(g,h), and the convolution result is subtracted from the denoised image to obtain b1, which can represent the difference between the denoised image and the ideal image.
[0069] Calculate the size of the error image and determine whether the error size is lower than the threshold. The error determination formula is as follows:
[0070]
[0071] If the error is lower than the threshold, it means that the difference between the denoised image and the ideal image is small, and the two can be considered equivalent.
[0072] If the error is higher than the threshold, the update step is entered, and the update formula is:
[0073]
[0074] r n represents the nth restored image, G * represents the flipping of the matrix G about the origin, C is a small constant added to prevent the denominator from being zero, and the error of the above steps is iterated to be lower than the threshold b n , then the ideal image is output.
[0075] When the sound pressure distribution G(g,h) is used as the blur kernel for convolution processing, it is described by the convolution of the sound pressure distribution function of the working plane of the focusing probe and the distribution of the acoustic impedance change of the measured plane:
[0076]
[0077] Among them, G(g,h) is the sound pressure distribution function, X(g,h) represents the distribution of acoustic impedance changes in the measured plane, N(g,h) represents the additive noise introduced by the system, and g and h represent the physical quantities of the actual distance;
[0078] In order to obtain an ideal unblurred two-dimensional image, it is necessary to deconvolve the equation, and the modeling result of G(g,h) is:
[0079]
[0080] Among them, f is the center frequency of the sound wave, a is the effective radius of the probe, c represents the speed of sound in the medium, l fIndicates the focal length of the probe.
[0081] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the protection scope of the present invention.
Claims
1. An ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration, characterized in that: The following steps are involved: Acquire a scanning image of an ultrasonic microscope and measurement data of a microprobe, wherein the measurement data includes a center frequency, a focal length, and an effective radius; Post-processing the ultrasonic microscope scanning image, wherein the post-processing includes normalization and grayscale processing; The center frequency of the microprobe is obtained by deconvolution of the hydrophone frequency response, the focal length is calculated according to the transit time and sound velocity of the maximum axial sound pressure, the effective radius is obtained according to the axial sound pressure distribution measurement, and the sound pressure distribution G(g,h) is determined by the measurement data; The image is denoised, and the sound pressure distribution G(g,h) is used as a blur kernel for convolution. The difference between the convolution result and the denoised image is taken, and the size of the error image is calculated to determine whether the error size is lower than the threshold. The error determination formula is as follows: If the error is lower than the threshold, it means that the difference between the denoised image and the ideal image is small, and the two can be considered equivalent; If the error is higher than the threshold, the update step is entered, and the update formula is: r n represents the nth restored image, G * represents the flipping of the matrix G about the origin, C is a small constant added to prevent the denominator from being zero, and the error of the above steps is iterated to be lower than the threshold b n , then the ideal image is output.
2. The ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration according to claim 1, characterized in that: The method for obtaining a scanning image of an ultrasonic microscope comprises: Adjust the distance between the probe of the ultrasonic microscope and the object to be measured so that the focusing plane of the sound beam of the probe coincides with the plane of the characteristic to be measured; Manually set the time range of the characteristic signal in the reflected echo as the gate threshold; Calculate the positive peak value / negative peak value / peak-to-peak value of the signal in the threshold as the grayscale value of the current position; The probe of the mobile ultrasonic microscope is scanned across the entire plane, and the above process is repeated to image the entire plane.
3. The ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration according to claim 1, characterized in that: The method for measuring the axial sound pressure distribution includes adjusting the relative positions of the needle hydrophone or the thin film hydrophone and the ultrasonic microprobe to ensure that the two are located on the same horizontal axis; and calculating the effective radius through the axial sound pressure distribution.
4. The ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration according to claim 3, characterized in that: Connect the synchronous signal output of the pulse transmitter and receiver to an oscilloscope, record the time difference between the synchronous signal and the pulse signal received by the hydrophone, i.e., the transit time, and calculate the focal length through the transit time; multiply the recorded transit time by the speed of sound in water at the test temperature to obtain the focal length of the probe; perform spectrum analysis on the signals collected by the electric pulse and the hydrophone to determine the center frequency of the probe, and calculate the center frequency of the probe after deconvolution using the frequency response data of the hydrophone.
5. The ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration according to claim 1, characterized in that: The method for denoising the image includes linearly mapping the pixel values collected by scanning to a range of 0 to 1, using a bilateral filter to smooth the image noise while retaining important edge information, and using f s Spatial domain function and f r The range function controls the balance between smoothing noise and preserving edges. The specific formula for the denoised image is: where Ω is a neighborhood centered at s, ||ss′|| is defined as the Euclidean distance between pixels s and s′, and |I(s)-I(s′)| represents the intensity difference between pixels s and s′; The spatial domain is: The range domain is: Adjust the neighborhood Ω of the spatial domain and the standard deviation σ of the range domain r Two parameters, the size of the neighborhood controls the degree of spatial domain smoothing, and the neighborhood is adjusted in the range of [2 to 5] pixels in steps of 1. A larger neighborhood will result in more spatial smoothing, but may blur details; The standard deviation controls the smoothness of the grayscale value intensity and is adjusted in steps of 0.1 in the range of [0-2]. A larger standard deviation makes the filter insensitive to brightness differences and loses brightness details, but can better remove noise, so that the additive noise of the original data can be effectively removed without blurring the texture details.
6. The ultrasonic microscopic image restoration method based on high-frequency focused probe acoustic field calibration according to claim 1, characterized in that: When the sound pressure distribution G(g,h) is used as the blur kernel for convolution processing, it is described by the convolution of the sound pressure distribution function of the working plane of the focusing probe and the distribution of the acoustic impedance change of the measured plane: Among them, G(g,h) is the sound pressure distribution function, X(g,h) represents the distribution of acoustic impedance changes in the measured plane, N(g,h) represents the additive noise introduced by the system, and g and h represent the physical quantities of the actual distance; In order to obtain an ideal unblurred two-dimensional image, it is necessary to deconvolve the equation, and the modeling result of G(g,h) is: Among them, f is the center frequency of the sound wave, a is the effective radius of the probe, c represents the speed of sound in the medium, l f Indicates the focal length of the probe.
Citation Information
Patent Citations
Real-time automatic focusing method for ultrasonic scanning microscope
CN113588798A
Full-scribing high-throughput color pathological imaging analysis instrument and method
CN115032196A
Frequency-space-time domain combined denoising and recovering system for low-signal-to-noise-ratio image sequence
CN115965552A
Computational methods and electronic camera apparatus for determining distance of objects, rapid autofocusing and obtaining improved focus images
EP0408224A2
Pseudo deconvolution method of recovering a distorted optical image
US6229649B1
Cited By
Target imaging method and system based on neighborhood sound field
CN120314963A
