Color blood flow laminar flow image optimization method, device, storage medium and electronic device

Abstract

The present invention provides a method, device, storage medium and electronic device for optimizing laminar flow images of color blood flow, wherein the method includes: performing envelope processing on the regulated signal; downsampling the envelope processing result; Gaussian blurring the downsampling result; upsampling the Gaussian blurring result; normalizing the upsampling result; performing longitudinal extreme point search on the normalized result; determining the coordinate positions of the upper and lower tube walls based on the extreme point search results; performing a single fitting on the coordinate positions of the upper and lower tube walls; determining the Gaussian kernel based on the single fitting result; performing time-delay correlation processing on the regulated signal; and determining the optimized filtered blood flow image based on the time-delay correlation processing result and the Gaussian kernel. Avoid changing the emission sequence of the blood flow to obtain a better ultrasonic blood flow image, reduce the burden of the system under multiplexed imaging and increase the robustness of the system in adjusting the signal performance; obtain the anisotropic morphology of the current blood flow through rapid calculation of a small amount of data.

Description

Technical Field

[0001] The present invention relates to the technical field of computer data processing, and in particular to a method, device, storage medium and electronic equipment for optimizing laminar flow images of color blood flow. Background Art

[0002] Ultrasonic blood flow imaging is a medical imaging technology that combines ultrasound imaging and the Doppler effect to visualize and evaluate the movement of blood flow in the human body. Among them, ultrasonic blood flow laminar flow is an ultrasonic blood flow imaging performed under specific presets, which usually occurs when the flow is at a relatively stable speed in the pipe. Laminar flow is characterized by less turbulence and eddy currents, allowing the fluid to flow in a more stable manner.

[0003] Ultrasound blood flow imaging is often used to detect and evaluate blood flow conditions in blood vessels. In this case, blood flow velocity and direction information is obtained through the Doppler effect. Blood flow imaging is often used in clinical practice to evaluate whether blood flow is normal, and laminar flow is a flow state that helps ensure that blood flow maintains relatively stable conditions during the evaluation process in order to obtain reliable ultrasound blood flow imaging data. Ultrasound blood flow imaging relies on the Doppler effect to measure blood flow velocity and direction, and laminar flow is a flow state that helps ensure that blood flow maintains relatively stable conditions during the evaluation process in order to obtain accurate ultrasound blood flow imaging data. Therefore, laminar flow is a fluid dynamic characteristic in ultrasound blood flow imaging, and the optimization of blood flow laminar flow images contributes to the accuracy and reliability of this technology. Summary of the invention

[0004] One of the purposes of the present invention is to provide a method, device, storage medium and electronic device for optimizing laminar flow images of color blood flow, which can avoid changing the emission sequence of blood flow and further obtain better ultrasonic blood flow images, reduce the burden of the system under multiplexed imaging and increase the robustness of the system in adjusting signal performance; the anisotropic morphology of the current blood flow can be obtained through rapid calculation of a small amount of data, and at the same time, combined with the background information of the current tissue, the visual optimization effect of the blood flow laminar flow can be better calculated.

[0005] An embodiment of the present invention provides a method for optimizing a color blood flow laminar flow image, comprising:

[0006] Performing envelope processing on the modulated signal;

[0007] Downsampling the envelope processing result;

[0008] Gaussian blur is performed on the downsampling result;

[0009] Upsample the Gaussian blur result;

[0010] Normalize the upsampling results;

[0011] Perform a longitudinal extreme point search on the normalized results;

[0012] Based on the extreme point search results, the coordinate positions of the upper and lower pipe walls are determined;

[0013] Fit the coordinate positions of the upper and lower tube walls once;

[0014] Based on the first fitting result, determine the Gaussian kernel;

[0015] Performing delay-related processing on the mediated signal;

[0016] Based on the time-delay correlation processing results and the Gaussian kernel, an optimized filtered blood flow image is determined.

[0017] Optionally, Gaussian blurring uses a one-dimensional Gaussian convolution kernel with a variance greater than a threshold to perform convolution in the depth direction. The expression of the one-dimensional Gaussian convolution kernel G(x) is:

[0018]

[0019] Among them, σ is the standard deviation of the current Gaussian distribution, e is a natural constant, and x is the position of the current point in the color blood flow image coordinates.

[0020] Optionally, the upsampling factor is equal to the downsampling factor.

[0021] Optionally, determining the coordinate positions of the upper and lower pipe walls based on the extreme point search results includes:

[0022] Perform peak sorting on the extreme point list in the extreme point search results, and take the top 2 results of each peak sorting to determine the coordinate positions of the upper and lower pipe walls;

[0023] Among them, for the extreme point search results, the coordinate points whose differences between different search results under the same sorting exceed three times the standard deviation of the statistical mean use the recalculated values ​​of n results. The recalculated value method first calculates the average value a and variance b of the n results, removes the values ​​outside the three times the variance as outliers, and then performs quadratic averaging to obtain the final coordinate position.

[0024] Optionally, a first fit uses a least squares regression method or a finely segmented least squares regression method, segmenting the blood vessel wall in g values ​​to obtain a slope array [K1, K2, ... Kp] for each interval, where K is the slope value, p is the number of segment groups for n longitudinal values, and g*p = n; and a slope interval mask GP with a scale of mxk is output according to different K values.

[0025] Optionally, determining the Gaussian kernel based on a first fitting result includes:

[0026] A two-dimensional anisotropic Gaussian kernel is obtained based on the slopes of p different longitudinal segment groups segmented from a single fitting result; wherein the probability distribution of the Gaussian kernel has different trends according to the K values ​​of different segments; the two-dimensional anisotropic Gaussian kernel is composed of the superposition of two one-dimensional Gaussian kernels.

[0027] Optionally, determining an optimized filtered blood flow image based on the delayed correlation processing result and the Gaussian kernel includes:

[0028] The fixed delay correlation V image in the delay correlation processing result is subjected to point-to-point anisotropic Gaussian convolution filtering Vg according to the GP mask; wherein each pixel point in V is convolved with an anisotropic Gaussian kernel;

[0029] The coordinates of the inter-group area where the slope of the groups [K1, K2, ... Kp] in the GP mask changes are recorded as an array [kx1, kx2 ... kx(p-1)], and the Vg values ​​in the coordinate list are averaged for the left and right pixels:

[0030] Vgkx1=0.5*Vg(kx1-1)+0.5*Vg(kx1+1);

[0031] Perform arc tangent calculation on the processed Vg image to obtain the optimized filtered blood flow image Vlm:

[0032] Vlm=arctan(Vgkx1).

[0033] An embodiment of the present invention provides a color blood flow laminar flow image optimization device, characterized by comprising:

[0034] An envelope processing module, used for performing envelope processing on the modulated signal;

[0035] A downsampling module, used for downsampling the envelope processing result;

[0036] Gaussian blur module, used to perform Gaussian blur on the downsampling result;

[0037] The upsampling module is used to upsample the Gaussian blur results;

[0038] A normalization module, used to normalize the upsampling results;

[0039] The extreme point search module is used to perform a longitudinal extreme point search on the normalized result;

[0040] A coordinate position determination module, used to determine the coordinate positions of the upper and lower pipe walls based on the extreme point search results;

[0041] A one-time fitting module is used to fit the coordinate positions of the upper and lower tube walls;

[0042] A Gaussian kernel determination module is used to determine the Gaussian kernel based on a fitting result;

[0043] A delay correlation processing module, used for performing delay correlation processing on the mediated signal;

[0044] The filtered blood flow image determination module is used to determine the optimized filtered blood flow image based on the time-delay correlation processing result and the Gaussian kernel.

[0045] An embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and a processor executes the computer program to implement any of the above methods.

[0046] An embodiment of the present invention provides an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any one of the methods described above.

[0047] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0048] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0050] Figure 1 It is a schematic diagram of a method for optimizing a color blood flow laminar flow image in an embodiment of the present invention;

[0051] Figure 2 It is a schematic diagram of slope fitting based on the carotid artery curve in an embodiment of the present invention;

[0052] Figure 3 A schematic diagram of establishing a two-dimensional Gaussian kernel by obtaining the aspect ratio of a Gaussian filter through a slope in an embodiment of the present invention;

[0053] Figure 4 It is a schematic diagram of a color blood flow laminar flow image optimization system in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0055] The basic working principles of ultrasonic blood laminar flow include:

[0056] 1. Ultrasound stimulation: First, the doctor or technician will use an ultrasound probe to send ultrasound waves to the area to be examined. These ultrasound waves are high-frequency sound waves that travel through the body without causing harmful radiation.

[0057] 2. Doppler effect: When ultrasound waves pass through blood vessels, they interact with the flowing blood. According to the Doppler effect, the sound waves are reflected back and their frequency (or wavelength) changes, which is related to the speed and direction of the blood.

[0058] 3. Frequency shift measurement: The ultrasound device detects this change in frequency, which is called frequency shift. The magnitude and direction of the frequency shift can provide information about the speed and direction of blood flow. If the blood moves toward the probe, the frequency shift will be positive; if the blood moves away from the probe, the frequency shift will be negative.

[0059] 4. Color Doppler: In order to visualize blood flow more clearly, modern ultrasound equipment usually uses color Doppler technology. This technology represents the blood flow speed and direction by superimposing colors on the image. Usually, red indicates blood flow toward the probe, blue indicates blood flow away from the probe, and the saturation of the color indicates the magnitude of the speed.

[0060] 5. Image generation: Ultrasound equipment converts the collected Doppler data into images that show the speed and direction of blood flow, which helps doctors diagnose various vascular diseases such as arterial stenosis, thrombosis, and valve abnormalities.

[0061] The embodiment of the present invention provides a method for optimizing laminar flow images of color blood flow, such as Figure 1 As shown, including:

[0062] S1, performing envelope processing on the mediated signal;

[0063] S2, downsampling the envelope processing result;

[0064] S3, performing Gaussian blurring on the downsampling result;

[0065] S4, upsampling the Gaussian blur result;

[0066] S5, normalizing the upsampling result;

[0067] S6, performing a longitudinal extreme point search on the normalized result;

[0068] S7. Determine the coordinate positions of the upper and lower pipe walls based on the extreme point search results;

[0069] S8, fitting the coordinate positions of the upper and lower tube walls;

[0070] S9, determining the Gaussian kernel based on the first fitting result;

[0071] S10, performing delay-related processing on the mediated signal;

[0072] S11. Determine an optimized filtered blood flow image based on the time-delay correlation processing result and the Gaussian kernel.

[0073] The present invention can avoid changing the emission sequence of blood flow and further obtain better ultrasonic blood flow images, reduce the burden of the system under multiplexed imaging and increase the robustness of the system in adjusting signal performance; the anisotropic morphology of the current blood flow can be obtained through rapid calculation of a small amount of data, and at the same time, combined with the background information of the current tissue, the visual optimization effect of blood flow laminar flow can be better calculated.

[0074] Specifically, Gaussian blurring uses a one-dimensional Gaussian convolution kernel with a variance greater than a threshold (the threshold can be 0.6) to perform convolution in the depth direction. The expression of the one-dimensional Gaussian convolution kernel G(x) is:

[0075]

[0076] Among them, σ is the standard deviation of the current Gaussian distribution, e is a natural constant, and x is the position of the current point in the color blood flow image coordinates.

[0077] Specifically, the upsampling multiple is equal to the downsampling multiple.

[0078] Specifically, determining the coordinate positions of the upper and lower pipe walls based on the extreme point search results includes:

[0079] The extreme point list in the extreme point search results is peak sorted, and the top 2 results of each peak sort are taken to determine the coordinate positions of the upper and lower pipe walls; the extreme point list is specifically, for example: assuming the point amplitude values ​​[A1, A2, A3…AN], horizontal coordinates [x1, x2, x3…xN] and vertical coordinates [y1, y2, y3…yN] obtained in the extreme point search. Re-sorting according to the amplitude value to [A3, A1, A3…AN], a new coordinate sorting [x3, x1, x2…xN] and vertical coordinates [y3, y1, y2…yN] can be obtained;

[0080] Among them, for the extreme point search results, for the coordinate points whose differences between different search results under the same ranking (for example, the same A ranking value, that is, (x8, y8) corresponding to A8) exceed three times the standard deviation of the statistical average, the recalculated values ​​are used using n results (assuming that the (x8, y8) coordinate exceeds the three times the standard deviation interval of the statistical average, the left and right n / 2 results, namely (x8-n / 2:x8+n / 2, y8-n / 2:y8+n / 2) are averaged). The recalculated value method first calculates the average value a and variance b of the n results, removes the values ​​outside the three times the variance as outliers, and then performs quadratic averaging to obtain the final coordinate position.

[0081] Specifically, Figure 2 As shown, a first fit uses the least squares regression method or the fine segmented least squares regression method, and the segmentation is divided into g values ​​to obtain the slope array [K1, K2, ... Kp] of the vascular wall in each interval, where K is the slope value, p is the number of segment groups for n longitudinal values, and g*p = n; and according to different K values, a slope interval mask GP with a scale of mxk is output.

[0082] Specifically, determining the Gaussian kernel based on a fitting result includes:

[0083] Based on the slopes of p different longitudinal segment groups separated from the first fitting result, a two-dimensional anisotropic Gaussian kernel (such as Figure 3 As shown, the horizontal and vertical proportional lengths of the Gaussian kernel can be obtained by using the horizontal and vertical projection direction components of the slope); wherein, the probability distribution of the Gaussian kernel has different trends according to the K value of different segments; the two-dimensional anisotropic Gaussian kernel is composed of calculating the superposition of two one-dimensional Gaussian kernels.

[0084] Specifically, the step of determining an optimized filtered blood flow image based on the delayed correlation processing result and the Gaussian kernel includes:

[0085] The fixed delay correlation V image in the delay correlation processing result is subjected to point-to-point anisotropic Gaussian convolution filtering Vg according to the GP mask; wherein each pixel point in V is convolved with an anisotropic Gaussian kernel;

[0086] The coordinates of the inter-group area where the slope of the groups [K1, K2, ... Kp] in the GP mask changes are recorded as an array [kx1, kx2 ... kx(p-1)], and the Vg values ​​in the coordinate list are averaged for the left and right pixels:

[0087] Vgkx1=0.5*Vg(kx1-1)+0.5*Vg(kx1+1);

[0088] Perform arc tangent calculation on the processed Vg image to obtain the optimized filtered blood flow image Vlm:

[0089] Vlm=arctan(Vgkx1).

[0090] The specific implementation steps of this application include the following:

[0091] Step 1: Perform delay correlation processing on the demodulated signal to obtain zero delay correlation (P) and fixed delay correlation (V) signals.

[0092] Step 2: Perform envelope processing on the demodulated signal to obtain an amplitude signal.

[0093] Step 3: Downsample the amplitude signal n times to obtain n downsampled 2D amplitude signal arrays, and perform Gaussian blurring on them. Gaussian blurring can use a one-dimensional Gaussian convolution kernel with a large variance to perform convolution in the depth direction. The expression of the one-dimensional Gaussian convolution kernel G(x) is:

[0094]

[0095] Among them, σ is the standard deviation of the current Gaussian distribution, e is a natural constant, and x is the position of the current point in the color blood flow image coordinates.

[0096] Step 4: Upsample the n fuzzified and downsampled 2D amplitude signal arrays, the upsampling multiple is equal to the downsampling multiple, and normalize all 2D amplitude signal arrays back to the original scale mxk. Perform a vertical extreme point search on the n scale-normalized signal arrays to obtain a list of n extreme value coordinates.

[0097] Step 5: sort the n extreme point lists in step 4 by peak value, and take the first two results of each peak sorting to obtain the coordinate positions of the upper and lower tube walls. Among them, for the coordinate points whose differences between different search results under the same sorting exceed three times the statistical average standard deviation, the recalculated values ​​of the n results can be used. The recalculated value method first calculates the average value a and variance b of the n results, removes the values ​​outside the three times variance as outliers, and then performs a secondary average to obtain the final coordinate position.

[0098] Step 6: Fit the coordinate positions of the upper and lower walls. The first fitting can use the least squares regression method or the more sophisticated segmented least squares regression method. With g values ​​as a segment, the slope array [K1, K2, ... Kp] of the vascular wall in each interval can be obtained, where K is the slope value, p is the number of segment groups for n longitudinal values, and g*p = n. And according to different K values, the slope interval mask (GP) with a scale of mxk is output.

[0099] Step 7: A two-dimensional anisotropic Gaussian kernel is obtained from the slopes of the p different longitudinal segment groups, and the probability distribution of the Gaussian kernel has different trends according to the K value of different segments. Among them, the two-dimensional anisotropic Gaussian kernel can be calculated by superposing two one-dimensional Gaussian kernels.

[0100] Step 8: Perform point-to-point anisotropic Gaussian convolution filtering (Vg) on ​​the fixed delay correlation (V) image obtained in step 1 according to the GP mask, wherein the anisotropic Gaussian kernel convolved with each pixel in V is calculated in step 7.

[0101] Step 9: Record the coordinates of the inter-group area where the slope of the groups [K1, K2, ... Kp] in the GP mask changes, and record them as an array [kx1, kx2 ... kx(p-1)], and average the Vg values ​​in the coordinate list for the left and right pixel points. The averaging method is:

[0102] Vgkx1=0.5*Vg(kx1-1)+0.5*Vg(kx1+1).

[0103] Step 10: Perform arctangent calculation on the Vg image processed in step 9 to obtain the optimized filtered blood flow image Vlm:

[0104] Vlm=arctan(Vgkx1).

[0105] The present application can avoid changing the emission sequence of blood flow and further obtain better ultrasonic blood flow images, reduce the burden of the system under multiplexed imaging and increase the robustness of the system in adjusting signal performance; the anisotropic morphology of the current blood flow can be obtained through rapid calculation of a small amount of data, and at the same time, combined with the background information of the current tissue, the visual optimization effect of blood flow laminar flow can be better calculated.

[0106] The embodiment of the present invention provides a color blood flow laminar flow image optimization device, such as Figure 4 As shown, including:

[0107] An envelope processing module 1 is used to perform envelope processing on the modulated signal;

[0108] Downsampling module 2, used for downsampling the envelope processing result;

[0109] Gaussian blur module 3, used for performing Gaussian blur on the downsampling result;

[0110] An upsampling module 4, used for upsampling the Gaussian blur result;

[0111] A normalization module 5, used for normalizing the upsampling result;

[0112] An extreme point search module 6 is used to perform a longitudinal extreme point search on the normalized result;

[0113] A coordinate position determination module 7, used to determine the coordinate positions of the upper and lower pipe walls based on the extreme point search results;

[0114] A primary fitting module 8 is used to perform a primary fitting on the coordinate positions of the upper and lower tube walls;

[0115] A Gaussian kernel determination module 9 is used to determine the Gaussian kernel based on a first fitting result;

[0116] The delay correlation processing module 10 is used to perform delay correlation processing on the mediated signal;

[0117] The filtered blood flow image determination module 11 is used to determine the optimized filtered blood flow image based on the delayed correlation processing result and the Gaussian kernel.

[0118] An embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. A processor executes the computer program to implement any of the above methods.

[0119] An embodiment of the present invention provides an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any one of the methods described above.

[0120] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for optimizing laminar flow images of color blood flow, characterized in that: include: Performing envelope processing on the modulated signal; Downsampling the envelope processing result; Gaussian blur is performed on the downsampling result; Upsample the Gaussian blur result; Normalize the upsampling results; Perform a longitudinal extreme point search on the normalized results; Based on the extreme point search results, the coordinate positions of the upper and lower pipe walls are determined; Fit the coordinate positions of the upper and lower tube walls once; Based on the first fitting result, determine the Gaussian kernel; Performing delay-related processing on the mediated signal; Based on the time-delay correlation processing results and the Gaussian kernel, an optimized filtered blood flow image is determined.

2. The color blood flow laminar flow image optimization method according to claim 1, characterized in that: Gaussian blurring uses a one-dimensional Gaussian convolution kernel with a variance greater than a threshold to perform convolution in the depth direction. The expression of the one-dimensional Gaussian convolution kernel G(x) is: Among them, σ is the standard deviation of the current Gaussian distribution, e is a natural constant, and x is the position of the current point in the image coordinates.

3. The color blood flow laminar flow image optimization method according to claim 1, characterized in that: The upsampling factor is equal to the downsampling factor.

4. The method for optimizing color blood flow laminar flow images according to claim 1, characterized in that: The step of determining the coordinate positions of the upper and lower pipe walls based on the extreme point search results includes: Perform peak sorting on the extreme point list in the extreme point search results, and take the top 2 results of each peak sorting to determine the coordinate positions of the upper and lower pipe walls; Among them, for the extreme point search results, the coordinate points whose differences between different search results under the same sorting exceed three times the standard deviation of the statistical mean use the recalculated values ​​of n results. The recalculated value method first calculates the average value a and variance b of the n results, removes the values ​​outside the three times the variance as outliers, and then performs quadratic averaging to obtain the final coordinate position.

5. The color blood flow laminar flow image optimization method according to claim 1, characterized in that: The first fitting uses the least squares regression method or the fine segmented least squares regression method, and the segmentation is divided into g values ​​to obtain the slope array [K1, K2, ... Kp] of the vascular wall in each interval, where K is the slope value, p is the number of segment groups for n longitudinal values, and g*p=n; and according to different K values, a slope interval mask GP with a scale of mxk is output.

6. The method for optimizing color blood flow laminar flow images according to claim 5, characterized in that: The step of determining the Gaussian kernel based on a first fitting result includes: A two-dimensional anisotropic Gaussian kernel is obtained based on the slopes of p different longitudinal segment groups segmented from a single fitting result; wherein the probability distribution of the Gaussian kernel has different trends according to the K values ​​of different segments; the two-dimensional anisotropic Gaussian kernel is composed of the superposition of two one-dimensional Gaussian kernels.

7. The method for optimizing color blood flow laminar flow images according to claim 5, characterized in that: The step of determining an optimized filtered blood flow image based on the delayed correlation processing result and the Gaussian kernel includes: The fixed delay correlation V image in the delay correlation processing result is subjected to point-to-point anisotropic Gaussian convolution filtering Vg according to the GP mask; wherein each pixel point in V is convolved with an anisotropic Gaussian kernel; The coordinates of the inter-group area where the slope of the groups [K1, K2, ... Kp] in the GP mask changes are recorded as an array [kx1, kx2 ... kx(p-1)], and the Vg values ​​in the coordinate list are averaged for the left and right pixels: Vgkx1=0.5*Vg(kx1-1)+0.5*Vg(kx1+1); Perform arc tangent calculation on the processed Vg image to obtain the optimized filtered blood flow image Vlm: Vlm=arctan(Vgkx1).

8. A color blood flow laminar flow image optimization device, characterized in that: include: An envelope processing module, used for performing envelope processing on the modulated signal; A downsampling module, used for downsampling the envelope processing result; Gaussian blur module, used to perform Gaussian blur on the downsampling result; The upsampling module is used to upsample the Gaussian blur results; A normalization module, used to normalize the upsampling results; The extreme point search module is used to perform a longitudinal extreme point search on the normalized result; A coordinate position determination module, used to determine the coordinate positions of the upper and lower pipe walls based on the extreme point search results; A one-time fitting module is used to fit the coordinate positions of the upper and lower tube walls; A Gaussian kernel determination module is used to determine the Gaussian kernel based on a fitting result; A delay correlation processing module, used for performing delay correlation processing on the mediated signal; The filtered blood flow image determination module is used to determine the optimized filtered blood flow image based on the time-delay correlation processing result and the Gaussian kernel.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the processor executes the computer program to implement the method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method according to any one of claims 1 to 7.

Images