An ultrasonic imaging method for internal defects of materials

Abstract

The application discloses a kind of super-resolution imaging methods for material internal defect ultrasonic detection, obtain ultrasonic image, the image is regionally segmented, obtain multiple small regions of N×N pixels, as original image, the selection of pixel size N is based on containing individual defect, to improve the timeliness of algorithm, only interpolation calculation is carried out to defect area. With the actual defect area representation proportion of original image as basis, determine the number of inserted pixel points, then the edge filling method is used to expand the edge of original image, select adjacent 4×4 image region as interpolation correlation area at each inserted pixel point, obtain the pixel value at the pixel point to be inserted by spatial coordinate mapping, correlation area determination and distance weight calculation, finally obtain super-resolution ultrasonic image, and then improve the resolution of original image. Compared with the cubic spline function interpolation method, the application simplifies the algorithm complexity and improves the imaging speed.

Description

Technical Field

[0001] This invention relates to the field of ultrasonic imaging, and in particular to a super-resolution imaging method for ultrasonic detection of internal defects in materials. Background Technology

[0002] Ultrasonic testing is an effective and widely used non-destructive testing method in industry. Visualization of ultrasonic testing data has become an important means of material or structural evaluation. Improving image resolution within the physical constraints of existing testing and imaging methods is of great value, and its significance for the quantitative assessment of internal material defects is self-evident. Therefore, with the rapid development of computer technology, the requirements for ultrasonic images are also increasing. The inherent limitations on resolution in existing ultrasonic imaging are an urgent need for modern ultrasonic testing technology.

[0003] Currently, methods for improving ultrasound image quality mainly fall into two categories: preprocessing techniques and post-processing techniques. Preprocessing techniques are related to the physical characteristics of the signals involved, such as coherence, bandwidth, nonlinear propagation, attenuation, and absorption. They improve and optimize signal generation or image acquisition, including spatial and frequency composite imaging, harmonic imaging, and pulse inversion. Post-processing techniques use image processing and machine learning methods (such as interpolation, Bayesian analysis, anti-aliasing, and deep learning) to enhance low-resolution images. Among these, image interpolation techniques have wide applications in image processing systems, enabling the acquisition of high-resolution images from a single low-resolution image.

[0004] There are numerous and constantly evolving image interpolation methods. Traditional interpolation algorithms include nearest neighbor interpolation, bilinear interpolation, and bicubic interpolation. Nearest neighbor interpolation assigns the pixel value corresponding to the insertion point to the nearest known pixel value. While computationally simple and fast, it is prone to jagged edges and mosaic effects. Bilinear interpolation determines the pixel weights based on the distances between the corresponding point and its four neighboring points, thus obtaining the target image pixel value. Compared to nearest neighbor interpolation, bilinear interpolation produces a smoother interpolation result and avoids mosaic effects, but it is prone to losing detail information in edge processing. Furthermore, for nearest neighbor and bilinear interpolation, their basis functions are direct interpolation basis functions, meaning that the interpolation coefficients can be simplified to pixel values. Bicubic interpolation uses a polynomial function to obtain the value of the pixel to be interpolated by weighted averaging of the adjacent sampling points around the interpolation point. It requires the use of a cubic polynomial interpolation function in both the horizontal and vertical directions. After interpolation, it can create smoother image edges and more delicate processing results, but the amount of computation increases and the processing time is longer. Summary of the Invention

[0005] To address the problems existing in current technologies, this invention proposes a super-resolution imaging method for ultrasonic detection of internal defects in materials. This invention designs a novel nonlinear interpolation function, simplifying the algorithm complexity and improving image imaging speed and quality. Furthermore, considering the interrelationships and influences between pixels, it integrates multiple pixels within a region for super-resolution image pixel calculation, resulting in superior performance. This method features a simple interpolation basis function structure, ease of implementation, and easy parameter adjustment, improving image detail and spatial resolution.

[0006] To achieve the above-mentioned objectives, this invention proposes a super-resolution imaging method for ultrasonic detection of internal defects in materials, comprising the following steps:

[0007] 1) Perform ultrasonic phased array testing on the material under test to obtain ultrasonic images of the area to be tested;

[0008] 2) Perform region segmentation on the ultrasound image of the area to be inspected to obtain a small N×N pixel region containing the defect, which serves as the original image;

[0009] 3) Based on the length and width of the defect image in the original image and the corresponding length and width of the actual defect area, determine the representation ratio of the original image to the actual defect area; use the representation ratio of the original image to the actual defect area to derive the representation ratio of the super-resolution image to the actual defect area.

[0010] 4) The edges of the original image are extended using the edge filling method to obtain an edge-extended image;

[0011] 5) Use a custom nonlinear interpolation function to perform image interpolation on the edge extension image. Map the pixel coordinates of the super-resolution image to the edge extension image through spatial coordinate transformation, obtain the pixel values ​​around the center point, and substitute the row and column distances of the coordinate points and the center point into the nonlinear interpolation function to form a weight matrix. Calculate each pixel point of the super-resolution image, and finally obtain the super-resolution ultrasound image.

[0012] Furthermore, the region segmentation in step 2) above includes the following steps:

[0013] 2.1) Convert the ultrasound image of the area to be detected into a grayscale image;

[0014] 2.2) Based on the distribution of background and defect pixel values ​​in the grayscale image, an intensity threshold is set to segment the background and defect information of the grayscale image to form a binarized image;

[0015] 2.3) Perform morphological closing operations on the binarized image to bridge narrow discontinuities and long grooves at image defects, thereby smoothing the defect contours;

[0016] 2.4) Obtain the coordinates of the boundary points of the defect contour in the smooth defect contour image;

[0017] 2.5) Calculate the centroid coordinates of the image defects based on the coordinates of the boundary points of the image contour;

[0018] 2.6) Based on the centroid coordinates of the image defects, select an imaging region of size N×N centered on the centroid coordinates as the original image.

[0019] Furthermore, in step 4) above, the boundary filling method selects the pixel values ​​of the edge rows and columns of the original image and uses the method of copying the outermost pixels to expand the boundary of the original image to form an edge-extended image.

[0020] Furthermore, the nonlinear interpolation function β(d) in step 5) above has the following expression:

[0021]

[0022] Where d is the independent variable of the function, which is the coordinate of the center point of the pixel to be inserted in the super-resolution image mapped onto the original image and the row and column distance of the surrounding pixels; k1 is the power, which takes the value of a positive integer; k2 and k3 are the waveform factors of the function, which are limited by the range of values ​​of d.

[0023] Furthermore, the spatial coordinate mapping in step 5) above specifically includes:

[0024] Assuming the coordinates of the pixel to be inserted in the super-resolution image are (i,j), the coordinates (x,y) of its center point on the original image are calculated using the following formula.

[0025]

[0026] Where m1 and m2 are the proportions of the defect region represented by the super-resolution image, m1 is the ratio of the number of rows in the super-resolution image to the number of rows in the original image, and m2 is the ratio of the number of columns in the super-resolution image to the number of columns in the original image.

[0027] Furthermore, the determination of the associated region in step 5) above specifically includes:

[0028] Calculate the pixel values ​​of the 4×4 image region surrounding the center point on the edge-extended image. The coordinates of the pixel values ​​in this region can be determined by four row coordinates (x1, x2, x3, x4) and four column coordinates (y1, y2, y3, y4). The calculation formula is as follows:

[0029]

[0030] Where [x] and [y] represent the floor function of rounding down the coordinates (x, y) of the center point on the original image, and l and p represent the number of rows and columns of the edge extension of the original image, respectively;

[0031] In the edge-extended image, the pixel values ​​of 16 coordinate points surrounding the center point are determined, and the pixel value matrix B is given by...

[0032]

[0033] Where f(x,y) represents the pixel value of the edge extension image, and the subscript indicates the corresponding row and column number.

[0034] Furthermore, the distance weight calculation in step 5) above specifically includes the following steps:

[0035] S5.1) Calculate the distance between the center of the pixel to be interpolated on the edge-extended image and the pixels of the surrounding 4×4 image region. This distance can be determined using four row distances (u1, u2, u3, u4) and four column distances (v1, v2, v3, v4). The formula for calculating this distance is as follows:

[0036]

[0037] Among them, the values ​​of u2, u3, v2, v3 are between 0 and 1, and the values ​​of u1, u4, v1, v4 are between 1 and 2;

[0038] S5.2) Substitute the values ​​of the four row distances (u1, u2, u3, u4) and the four column distances (v1, v2, v3, v4) into the nonlinear interpolation function to obtain the weight matrices A and C of the distance between the 4×4 image region and the center point to be interpolated. Their expressions are as follows:

[0039]

[0040] Furthermore, step 5) above, calculating each pixel of the super-resolution image, includes the following steps:

[0041] Calculate the pixel value g(i,j) of each pixel to be inserted in the super-resolution image:

[0042] g(i,j)=ABC

[0043] The pixel values ​​g(i,j) of the pixels to be inserted are inserted into the original image to form a super-resolution ultrasound image.

[0044] The beneficial effects of this invention are:

[0045] A nonlinear interpolation function was designed to simplify the interpolation coefficients to pixel values, solving the problem of estimating interpolation coefficients when using higher-order B-splines as interpolation basis functions, thus simplifying the algorithm complexity. By segmenting the original image into regions, defect region images are obtained. The size of the defect region image is generally less than one-fifth of the original image, reducing image processing load and improving imaging speed.

[0046] The interpolation basis function proposed in this invention has the characteristics of simple structure, easy implementation and easy parameter adjustment. When applied to the ultrasound image super-resolution algorithm, it can improve the image details and spatial resolution. Attached Figure Description

[0047] Figure 1 This is a flowchart of the method in an embodiment of the present invention;

[0048] Figure 2 The ultrasound image obtained in the embodiments of the present invention;

[0049] Figure 3 This is an ultrasound image showing the marked defect contour and centroid coordinates in an embodiment of the present invention. In the figure, 1 represents the edge contour of the image defect; 2 represents the centroid coordinates of the image defect;

[0050] Figure 4 The images shown are the original images of the defective regions in this embodiment of the invention; (a) defect 1; (b) defect 2.

[0051] Figure 5 This is an image of the nonlinear interpolation function in an embodiment of the present invention;

[0052] Figure 6 This is a schematic diagram illustrating the nonlinear interpolation principle in an embodiment of the present invention;

[0053] Figure 7 These are super-resolution ultrasound images from embodiments of the present invention. (a) Defect 1; (b) Defect 2 Detailed Implementation

[0054] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0055] like Figure 1 The diagram shown is a flowchart of the method of the present invention, which specifically includes the following steps:

[0056] 1) Perform ultrasonic phased array testing on the material under test to obtain ultrasonic images of the area to be tested;

[0057] In a specific embodiment of the present invention, such as Figure 2 As shown, ultrasonic phased array testing was performed on a plate sample containing through-hole defects. The phased array probe used a 1×16 linear array with an angle scanning range of n1-n2, where n1 = 25° and n2 = 75°, to acquire color ultrasonic images.

[0058] 2) Perform region segmentation on the ultrasound image of the area to be inspected to obtain a small N×N pixel region containing the defect, which serves as the original image;

[0059] 3) Based on the length and width of the defect image in the original image and the corresponding length and width of the actual defect area, determine the representation ratio of the original image to the actual defect area; use the representation ratio of the original image to the actual defect area to derive the representation ratio of the super-resolution image to the actual defect area.

[0060] 4) The edges of the original image are extended using the edge filling method to obtain an edge-extended image;

[0061] 5) Use a custom nonlinear interpolation function to perform image interpolation on the edge extension image. Map the pixel coordinates of the super-resolution image to the edge extension image through spatial coordinate transformation, obtain the pixel values ​​around the center point, and substitute the row and column distances of the coordinate points and the center point into the nonlinear interpolation function to form a weight matrix. Calculate each pixel point of the super-resolution image, and finally obtain the super-resolution ultrasound image.

[0062] As a preferred embodiment of the present invention, the region segmentation in step 2) includes the following steps:

[0063] 2.1) Convert the ultrasound image of the area to be detected into a grayscale image;

[0064] 2.2) Based on the distribution of background and defect pixel values ​​in the grayscale image, an intensity threshold is set to segment the background and defect information of the grayscale image to form a binarized image;

[0065] 2.3) Perform morphological closing operations on the binarized image to bridge narrow discontinuities and long grooves at image defects, thereby smoothing the defect contours;

[0066] 2.4) As Figure 3 As shown, obtain the coordinates of the boundary points of the defect contour in the smooth defect contour image;

[0067] 2.5) Calculate the centroid coordinates of the image defects based on the coordinates of the boundary points of the image contour;

[0068] 2.6) For example Figure 4 As shown, based on the centroid coordinates of the image defect, an imaging region of size N×N centered at the centroid coordinates is selected as the original image. In a specific embodiment of the present invention, N = 100.

[0069] As a preferred embodiment of the present invention, in step 4), the boundary filling method selects the pixel values ​​of the edge rows and columns of the original image and uses the method of copying the outermost pixels to expand the boundary of the original image to form an edge-extended image.

[0070] As a preferred embodiment of the present invention, such as Figure 5 As shown, the nonlinear interpolation function β(d) in step 5) has the following expression:

[0071]

[0072] Where d is the independent variable of the function, which is the coordinates of the center point of the pixel to be inserted in the super-resolution image mapped onto the original image and the row and column distance of the surrounding pixels; k1 is the power, which takes the value of a positive integer; k2 and k3 are function waveform factors, which are limited by the range of d. In a specific embodiment of the present invention, the power k1 is preferably 3, and the function waveform factors k2 and k3 are preferably in the range of 0.5 to 0.8.

[0073] In a preferred embodiment of the present invention, the spatial coordinate mapping in step 5) above specifically includes:

[0074] Assuming the coordinates of the pixel to be inserted in the super-resolution image are (i,j), the coordinates (x,y) of its center point on the original image are calculated using the following formula.

[0075]

[0076] Where m1 and m2 are the proportions of the defect region represented by the super-resolution image, m1 is the ratio of the number of rows in the super-resolution image to the number of rows in the original image, and m2 is the ratio of the number of columns in the super-resolution image to the number of columns in the original image. In a specific embodiment of the present invention, m1 = m2 = 2.

[0077] In a preferred embodiment of the present invention, the determination of the associated region in step 5) above specifically includes:

[0078] Calculate the pixel values ​​of the 4×4 image region surrounding the center point on the edge-extended image. The coordinates of the pixel values ​​in this region can be determined by four row coordinates (x1, x2, x3, x4) and four column coordinates (y1, y2, y3, y4). The calculation formula is as follows:

[0079]

[0080] Where [x], [y] represent the rounding down of the center point coordinates (x, y) on the original image, and l and p represent the number of rows and columns of the edge extension of the original image, respectively; as a preferred embodiment of the present invention, l = p = 2 is selected.

[0081] In the edge-extended image, the pixel values ​​of 16 coordinate points surrounding the center point are determined, and the pixel value matrix B is given by...

[0082]

[0083] Where f(x,y) represents the pixel value of the edge extension image, and the subscript indicates the corresponding row and column number.

[0084] In a preferred embodiment of the present invention, step 5) of the distance weight calculation specifically includes the following steps:

[0085] S5.1) Calculate the distance between the center of the pixel to be interpolated on the edge-extended image and the pixels of the surrounding 4×4 image region. This distance can be determined using four row distances (u1, u2, u3, u4) and four column distances (v1, v2, v3, v4). The formula for calculating this distance is as follows:

[0086]

[0087] Among them, the values ​​of u2, u3, v2, v3 are between 0 and 1, and the values ​​of u1, u4, v1, v4 are between 1 and 2;

[0088] S5.2) Substitute the values ​​of the four row distances (u1, u2, u3, u4) and the four column distances (v1, v2, v3, v4) into the nonlinear interpolation function to obtain the weight matrices A and C of the distance between the 4×4 image region and the center point to be interpolated. Their expressions are as follows:

[0089]

[0090] In a preferred embodiment of the present invention, step 5) of calculating each pixel of the super-resolution image includes the following steps:

[0091] like Figure 6 As shown, calculate the pixel value g(i,j) of each pixel to be inserted in the super-resolution image:

[0092] g(i,j)=ABC

[0093] like Figure 7 As shown, the pixel values ​​g(i,j) of the pixels to be inserted are inserted into the original image to form a super-resolution ultrasound image.

[0094] To evaluate the resolution improvement performance of the interpolated image, image pixels per unit area, peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) were used as evaluation metrics to measure the image reconstruction quality. The formula for calculating PSNR is as follows:

[0095]

[0096] In the formula, n is the number of bits per pixel, and MSE represents the mean square error between the original image and the reconstructed image.

[0097] The formula for calculating SSIM is:

[0098]

[0099] In the formula, Let u represent the variances of images X and Y, respectively. X u YLet σ represent the mean of image X and Y, respectively. XY Let C1 and C2 represent the covariance of images X and Y, respectively, and C1 and C2 are constants.

[0100] After super-resolution processing of the original image, its resolution was increased from 100×100 to 200×200, improving the image's pixel representation of defects in that region. Two sets of defect images were evaluated: Defect 1, PSNR = 31.6599dB, SSIM = 0.9684; Defect 2, PSNR = 30.1426dB, SSIM = 0.9589.

[0101] The detailed descriptions listed above are merely specific descriptions of feasible embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. All equivalent embodiments or modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention.

Claims

1. A super-resolution imaging method for ultrasonic detection of internal defects in materials, characterized in that, Includes the following steps: 1) Perform ultrasonic phased array testing on the material under test to obtain ultrasonic images of the area to be tested; 2) Perform region segmentation on the ultrasound image of the area to be detected to obtain... Pixels contain small areas of defects, which are used as the original image; 3) Based on the length and width of the defect image in the original image and the length and width of the corresponding actual defect area, determine the representation ratio of the original image to the actual defect area; use the representation ratio of the original image to the actual defect area to derive the representation ratio of the super-resolution image to the actual defect area. 4) The edges of the original image are extended using the edge filling method to obtain an edge-extended image; 5) Use a custom nonlinear interpolation function to perform image interpolation on the edge extension image. Map the pixel coordinates of the super-resolution image to the edge extension image through spatial coordinate transformation, obtain the pixel values ​​around the center point, and substitute the row and column distances between the coordinate points and the center point into the nonlinear interpolation function to form a weight matrix. Calculate each pixel point of the super-resolution image, and finally obtain the super-resolution ultrasound image. The nonlinear interpolation function in step 5) Its function expression is, ； in, is the function's independent variable, representing the coordinates of the center point of the super-resolution image to be inserted onto the original image and the row and column distances of the surrounding pixels; It represents a power and takes the value of a positive integer. and The waveform factor of the function is determined by... The range of values ​​is limited.

2. The super-resolution imaging method for ultrasonic detection of internal defects in materials according to claim 1, characterized in that, The region segmentation in step 2) includes the following steps: 2.1) Convert the ultrasound image of the area to be detected into a grayscale image; 2.2) Based on the distribution of background and defect pixel values ​​in the grayscale image, an intensity threshold is set to segment the background and defect information of the grayscale image to form a binarized image; 2.3) Perform morphological closing operations on the binarized image to bridge the discontinuities and narrow grooves at the defects in the image, thereby smoothing the defect contours; 2.4) Obtain the coordinates of the boundary points of the defect contour in the smooth defect contour image; 2.5) Calculate the centroid coordinates of the image defects based on the coordinates of the boundary points of the image contour; 2.6) Based on the centroid coordinates of the image defect, select a center point centered on the centroid coordinates. The size of the imaging region is used as the original image.

3. The super-resolution imaging method for ultrasonic detection of internal defects in materials according to claim 1, characterized in that, In step 4), the boundary filling method selects the pixel values ​​of the edge rows and columns of the original image and uses the method of copying the outermost pixels to expand the boundary of the original image to form an edge-extended image.

4. The super-resolution imaging method for ultrasonic detection of internal defects in materials according to claim 1, characterized in that, The spatial coordinate mapping in step 5) specifically includes: Assume the coordinates of the pixel to be inserted in the super-resolution image are Its coordinates are mapped to the center point on the original image. The calculation is performed using the following formula. ； in and The proportion of the defect region represented by the super-resolution image. This is the ratio of the number of rows in the super-resolution image to the number of rows in the original image. This is the ratio of the number of columns in the super-resolution image to the number of columns in the original image.

5. The super-resolution imaging method for ultrasonic detection of internal defects in materials according to claim 1, characterized in that, The determination of the associated region in step 5) specifically includes: Calculate the surrounding area, including the center point, mapped onto the edge-extended image. The pixel values ​​of an image region, whose coordinates can be represented by four row coordinates. and 4 column coordinates Confirmed, the calculation formula is as follows: ； in, , Represents the coordinates of the center point on the original image. Round the value down to the nearest integer. and These represent the number of rows and columns of the edge extension in the original image, respectively; In the edge-extended image, determine the pixel values ​​of 16 coordinate points surrounding the center point, and their pixel value matrix. for, ； in, The image pixel values ​​are extended to the edges, and the subscripts indicate the corresponding row and column numbers.

6. The super-resolution imaging method for ultrasonic detection of internal defects in materials according to claim 1, characterized in that, Step 5) specifically includes the following steps for calculating the distance weight: S5.1) Calculate the distance from the center of the pixel to be interpolated on the edge-extended image to the surrounding area. The distance between pixels in an image region can be represented by four row distances. Distance to 4 columns The distance calculation formula is as follows: ； in, The value is located at between, The value is located at between; S5.2) Distance of 4 rows Distance to 4 columns Substituting the numerical values ​​into the nonlinear interpolation function, we obtain... Weight matrix of distances between image region and center point to be interpolated and Its expression is, 。 7. The super-resolution imaging method for ultrasonic detection of internal defects in materials according to claim 1, characterized in that, Step 5) involves calculating each pixel of the super-resolution image, which includes the following steps: Calculate the pixel value of each pixel to be inserted in the super-resolution image. : ； The pixel value of the pixel to be inserted Insert it into the original image to create a super-resolution ultrasound image.

Images