A non-destructive testing method based on spatiotemporal information fusion of infrared thermal imaging

Abstract

This invention discloses a non-destructive testing method based on spatiotemporal information fusion of infrared thermal imaging, belonging to the field of non-destructive testing technology. The method includes: constructing a linear laser heat source scanning infrared thermal imaging non-destructive testing system; acquiring dynamic infrared thermal image sequences; calculating and analyzing the dynamic infrared thermal image sequence data; and processing the acquired dynamic infrared thermal image sequences using a post-processing method based on spatiotemporal information fusion of infrared thermal imaging to obtain a full-field thermal response image with surface crack defects. This invention not only fuses spatiotemporal information to obtain higher resolution data and improves computational efficiency while obtaining more comprehensive and accurate full-field thermal response characteristics, but also has better flexibility and wider applicability, achieving the goal of non-destructive testing of metal surface defects.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing technology, specifically to a nondestructive testing method that fuses spatiotemporal information from infrared thermal imaging. Background Technology

[0002] Non-destructive testing (NDT) technology has emerged as an indispensable and effective tool alongside industrial development. It enables the effective evaluation of the internal structure and performance of materials or components without damaging them, utilizing physical and chemical methods to ensure their safety and reliability. Currently, widely used NDT methods include radiographic testing, ultrasonic testing, liquid penetrant testing, magnetic particle testing, eddy current testing, acoustic emission testing, and infrared thermal imaging.

[0003] Infrared thermography is an advanced non-destructive testing technique that uses an infrared camera to capture the thermal radiation from an object's surface and convert it into an image of its temperature distribution, revealing the object's surface or internal microstructure, defects, and thermal properties. As a classic non-contact testing method, it offers advantages such as speed, accuracy, convenience, and intuitiveness, and has gained widespread attention and application in recent years.

[0004] Existing infrared thermal image post-processing methods face numerous challenges, including: uneven heating of the sample surface, low emissivity, lateral thermal diffusion, limited spatial resolution and image acquisition frequency of thermal imagers, and interference from environmental noise. These factors lead to low contrast and high noise in infrared images, thus making defect detection difficult.

[0005] Existing research on processing dynamic infrared thermal image sequences suffers from the following shortcomings: 1. Limited data dimensionality: A single dimension may not provide sufficient contrast and edge information, potentially leading to the loss of important information such as temperature change trends in spatial or temporal dimensions, thus limiting the ability to comprehensively analyze complex thermal phenomena; 2. Low detection sensitivity and poor adaptability: Existing algorithms are significantly affected by the lateral diffusion of heat waves when processing dynamic infrared thermal images, impacting detection accuracy and making it difficult to detect small or deep defects. Furthermore, the use of some thermal image post-processing algorithms is limited by specific types of thermal excitation sources; 3. Low detection efficiency: Some traditional methods suffer from slow detection speeds when processing infrared thermal image sequences, limiting their application in real-time or rapid detection scenarios. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a non-destructive testing method based on the spatiotemporal information fusion of infrared thermal imaging. By establishing a mathematical model, information from the spatial and temporal dimensions is fused to obtain higher resolution data, thereby improving computational efficiency and obtaining more comprehensive and accurate full-field thermal response characteristics of defects and damage.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A non-destructive testing method based on spatiotemporal information fusion of infrared thermal imaging includes:

[0009] A fast-moving linear laser is used as a thermal excitation source to scan and excite the surface of a metal specimen. A thermal infrared acquisition device is used to acquire a dynamic infrared thermal image sequence containing defect damage information within the detection area. The dynamic infrared thermal image sequence is processed using a spatiotemporal information fusion post-processing method to obtain a full-field thermal response image with surface crack defects.

[0010] Compared with existing technologies, the present invention has the following beneficial effects:

[0011] First, this method integrates multi-dimensional data. While rapidly identifying the location and shape of cracks, it also assesses crack development and the thermal diffusion characteristics of the material, enabling a more comprehensive understanding and evaluation of the material's thermal properties and structural integrity.

[0012] Second, this method offers good flexibility. When calculating between two adjacent thermal images, it can automatically eliminate the influence of linear laser heat sources on the detection. Therefore, this method is not limited to a specific thermal excitation source and can combine different types of thermal excitation methods to acquire and analyze thermal image data.

[0013] Third, this method is beneficial for improving the signal-to-noise ratio. Spatiotemporal information fusion can effectively separate and suppress noise, improving the signal-to-noise ratio, which is more conducive to the inspection of smaller and deeper defects. This study successfully detected surface crack defects with a width of 3 μm and above.

[0014] Fourth, this method has a wider range of applications. It is not only suitable for detecting cracks on planar surfaces, but also for detecting cracks and pitting defects on gear meshing surfaces, capable of detecting cracks as small as 0.01 mm. 2 Pitting defects on the worn area. Attached Figure Description

[0015] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0016] Figure 1 This is a flowchart illustrating the operation of a non-destructive testing method for infrared thermal imaging spatiotemporal information fusion according to an embodiment of the present invention.

[0017] Figure 2 This is a flowchart illustrating the calculation process of a non-destructive testing method based on spatiotemporal information fusion of infrared thermal imaging, according to an embodiment of the present invention.

[0018] Figure 3 This is a schematic diagram of a metal specimen containing surface crack defects detected in an embodiment of the present invention;

[0019] Figure 4 This is a schematic diagram of a non-destructive testing system for linear laser thermal source scanning infrared thermal imaging of metal surface crack defects in an embodiment of the present invention.

[0020] Figure 5 The following is the result of post-processing the infrared thermal image sequence acquired by the thermal imager in the embodiments of the present invention; wherein, (a) is the original thermal image, (b) is the original thermal image minus the background image, (c) is the calculation result of a set of two adjacent thermal images using the post-processing method of infrared thermal imaging spatiotemporal information fusion, (d) is the cumulative calculation of each set of results, and (e) is the full-field thermal response image with surface crack defects. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of the invention described below can be combined with each other as long as they do not conflict with each other. The accompanying drawings, which are provided to further understand the invention and constitute a part of this invention, illustrate exemplary embodiments of the invention and their descriptions, and are used to explain the invention, but do not constitute an undue limitation of the invention.

[0022] like Figure 1 The illustration shows a specific embodiment of a non-destructive testing method based on spatiotemporal information fusion of infrared thermal imaging provided by the present invention. This method is applicable to linear laser thermal source scanning infrared thermal imaging non-destructive testing for detecting cracks and defects on metal surfaces. This embodiment includes the following steps:

[0023] Step 10: Build a linear laser thermal source scanning infrared thermal imaging non-destructive testing system, including: setting up a thermal excitation device, optimizing the design of linear laser thermal excitation loading parameters, setting up a thermal infrared acquisition device, installing a suitable microscope head and setting appropriate temperature measurement range, acquisition frame rate and other parameters.

[0024] Step 11: Acquire dynamic infrared thermal image sequence, including: use an infrared thermal imager to acquire temperature images of the surface of the tested component, and ensure that the laser scanning and infrared imaging are started and stopped synchronously in time.

[0025] Step 12: Calculate and analyze dynamic infrared thermal image sequence data, including: process the acquired dynamic infrared thermal image sequence using a post-processing method of infrared thermal imaging spatiotemporal information fusion to obtain a full-field thermal response image with surface crack defects.

[0026] In step 10 above, the construction of a linear laser thermal source scanning infrared thermal imaging non-destructive testing system is further described as follows: Optimize the use of a laser marking machine by connecting it to a computer, opening the laser marking machine control software, manually adjusting the laser focus to form a uniform 10 µm laser spot on the focal plane to irradiate the component surface, using a linear array to draw the laser movement scanning path, setting the laser line thermal source width to 2 mm, the scanning speed to approximately 4 mm / s, and the laser power to 5 W, and using red light to check the marking position before each "engraving"; use a long-wavelength (7.5-14 µm) infrared camera (FLUKE RSE60) with a maximum resolution of 640 × 480 pixels, attaching an infrared microscope head to the infrared camera and fixing it on a tripod to ensure it is perpendicularly aligned with the surface of the component being tested, connecting it to the computer, opening the thermal imager's thermal analysis software, selecting a temperature range of 0~650℃, setting the acquisition frame rate to 30 Hz, and adjusting the tripod height so that the objective lens distance is close to the minimum focusing distance of the infrared microscope head (15 cm).

[0027] In step 11 above, an infrared thermal imager is used to acquire temperature images of the surface of the component under test. Further, the following steps are taken: the “Record” function is clicked in the thermal image analysis software to enable the infrared camera to capture the thermal response of the sample surface. At the same time as clicking “Record”, laser scanning is started. After “Record” is completed, the imaging area is re-selected and saved frame by frame in CSV and BMP formats to obtain a dynamic thermal image sequence of the component surface temperature changing over time.

[0028] In step 12 above, a full-field thermal response image with surface crack defects is obtained. Further, the following steps are performed: the acquired dynamic infrared thermal image sequence is calculated and analyzed using the post-processing method of infrared thermal imaging spatiotemporal information fusion in this invention. The background image is subtracted frame by frame from the dynamic thermal image sequence containing multiple frames. The mathematical model established in this invention is used for calculation, where the time step corresponding to the frame rate of the thermal imager is input to calculate the time derivative. The optical flow method is used to solve for the velocity vectors of heat flow along the x-axis and y-axis directions. and We obtained full-field thermal response images with surface crack defects.

[0029] like Figure 2The calculation process of the infrared thermal imaging spatiotemporal information fusion method shown is as follows: First, data input: input the original thermal image data, including the background image and the dynamic thermal sequence under line laser excitation, and input the total number of frames of the dynamic thermal sequence and the frame rate of the thermal imager. Second, perform multi-step calculation: perform background subtraction and Gaussian filtering on the dynamic thermal sequence frame by frame to eliminate background noise and smooth the image; calculate the spatial row gradient and spatial column gradient of temperature pixel by pixel for each frame of thermal image; calculate the temporal gradient of temperature pixel by pixel for two adjacent frames, and use the optical flow method to solve the heat flow velocity in the x and y directions, and assign the calculation result to the previous frame in the adjacent frame. Then, fuse the spatiotemporal information pixel by pixel based on formula (15). Finally, output the non-destructive detection result image of the defect. This process significantly improves the detection spatial resolution of weak defects in dynamic thermal signals through spatiotemporal feature coupling.

[0030] The full-field thermal response image with surface crack defects is obtained as follows:

[0031] The background image without thermal excitation source is subtracted frame by frame from the dynamic thermal image sequence containing multiple frames to obtain a dynamic thermal image sequence containing multiple frames with background noise removed. The mathematical model of spatiotemporal information fusion of infrared thermal imaging established in this invention (Equations (15) and (16) below) is used for calculation, wherein the time step corresponding to the frame rate of the thermal imager is input to calculate the time derivative, and the optical flow method is used to solve the velocity vector of heat flow along the x-axis and y-axis. and We obtained full-field thermal response images with surface crack defects.

[0032] Specifically, obtaining full-field thermal response images with surface crack defects includes:

[0033] In the acquired dynamic thermal image sequence, the thermal image temperature field is represented as a function related to spatial location and time:

[0034] (1)

[0035] In the formula, Let be the temperature value of the feature parameter pixel (x, y) at time t, where x and y represent the spatial position of the heat source at time t.

[0036] The derivative of the temperature field in the thermal image with respect to time is calculated using the following formula:

[0037] (2)

[0038] In the formula, the Lagrange derivative is... Euler's derivative represents the rate of change of temperature over time. This represents the rate of temperature change over time at a fixed pixel location. and These represent the contributions of temperature changes caused by the movement of the heat source in the x and y directions to the total rate of temperature change, respectively. This indicates the contribution of temperature changes caused by factors such as heat conduction, heat convection, or heat radiation to the total rate of temperature change in the absence of a moving heat source. and These represent the velocity vectors of heat flow along the x-axis and y-axis, respectively.

[0039] The velocity vectors of heat flow along the x-axis and y-axis can be determined using the optical flow method. and Specifically as follows:

[0040] Assuming constant brightness, a feature parameter pixel value Within time dt, the pixel values ​​of the feature parameters move a distance of (dx, dy) to the next frame. The spatial changes of these pixel values ​​are continuous, and the temperature distribution does not change drastically. The formula is as follows:

[0041] (3)

[0042] In the formula, Let dx be the temperature value of the feature parameter pixel (x, y) at time t, dy be the displacement in the x direction, dt be the displacement in the y direction, and dt be the time increment.

[0043] Based on the assumptions of small motion and displacement, continuity, and differentiability, it can be known that the propagation and diffusion of heat between adjacent frames is continuous, smooth, and differentiable. A Taylor expansion of equation (3) yields the following formula:

[0044] (4)

[0045] In the formula, This represents a second-order infinitesimal, which can be neglected. Combining equations (3) and (4), we get:

[0046] (5)

[0047] As can be seen from the locality assumption, local thermal diffusion characteristics can be captured by calculating the heat flow velocity vector of each pixel or small region.

[0048] By minimizing a global energy function that includes data terms and a smoothing term to add additional constraints, the Horn-Schunck algorithm is used to solve for the energy along the path. Axial direction and The velocity vector of heat flow in the axial direction and .

[0049] Specifically, the background image is subtracted frame by frame from the dynamic thermal image sequence containing multiple frames. Further, the temperature field without thermal excitation is subtracted from the temperature field at each moment in the dynamic thermal image sequence to obtain a dynamic thermal image sequence containing multiple frames with background noise removed.

[0050] Specifically, the data processing is further performed using the post-processing method for establishing spatiotemporal information fusion of infrared thermal imaging as described in this invention:

[0051] For the above dynamic thermal sequence, where each frame of thermal image is M×N in size, the matrix expression is:

[0052] (6)

[0053] In the formula, T represents the acquired raw infrared thermal image, i represents the frame index in the dynamic thermal sequence, and L... t This indicates the number of frames contained in the dynamic thermal sequence, 'a' represents the element in the original infrared thermal image, and 'M' and 'N' represent the number of images, rows, and columns, respectively.

[0054] Calculate the spatial temperature gradient for all pixels in each frame of the thermal image. Calculate the row gradient for each row of a single frame of the thermal image. :

[0055] (7)

[0056] In the formula, k and j represent the row index and column index of each frame of thermal image. and These represent the temperature values ​​in the j-th row and (k+1)-th column of the i-th frame of the thermal image, respectively.

[0057] The matrix expression of equation (7) is:

[0058] (8)

[0059] In the formula, This represents an element in the row gradient matrix.

[0060] The row gradient is obtained by calculating the temperature gradient of each column in a single frame of a thermal image, using the following formula:

[0061] (9)

[0062] In the formula, and These represent the temperature values ​​in the k-th column and (j+1)-th row of the i-th frame of the thermal image, respectively.

[0063] The matrix expression of equation (9) is:

[0064] (10)

[0065] In the formula, This represents an element in the column gradient matrix.

[0066] Calculate the time-temperature gradient for all pixels in two adjacent thermal images. The formula is as follows:

[0067] (11)

[0068] In the formula, This represents the sampling time interval of the thermal imager, which is inversely proportional to the imager's frame rate.

[0069] The matrix expression of equation (11) is:

[0070] (12)

[0071] In the formula, This represents an element in the time-temperature gradient matrix.

[0072] Calculate the path along the edge using the optical flow method for two adjacent thermal images. Axial direction and The velocity vector of heat flow in the axial direction The formula is as follows:

[0073] (13)

[0074] In the formula, and These represent the horizontal components of the heat flow velocity. and vertical components α represents the regularization parameter, used to balance the data terms and the smoothing term; Horn-Schunck represents the Horn-Schunck algorithm. β is the number of iterations in the algorithm.

[0075] The matrix expression of equation (13) is:

[0076] (14)

[0077] In the formula, and These represent the elements in the velocity vector matrix of heat flow along the x-axis and y-axis, respectively.

[0078] The formula for calculating the final defect detection result by fusing temporal and spatial dimensional information is as follows:

[0079] (15)

[0080] The matrix of equation (15) The expression is:

[0081] (16)

[0082] like Figure 3 As shown, the experimental specimen was made by mirror polishing the surface of a stainless steel specimen with a length of 20 mm, a width of 10 mm, and a height of 10 mm to achieve a suitable surface roughness requirement. The specimen was then cut by wire cutting, and finally the two parts were overlapped along the cut surface, so that the surface of the specimen contained cracks with a width of 3 μm or more.

[0083] like Figure 4 As shown, the laser line scanning infrared thermal imaging non-destructive testing system constructed in this invention mainly comprises three parts: a thermal excitation device (linear laser heat source) capable of laser line scanning, a thermal infrared acquisition device (metal component, infrared thermal imager, laser marking machine, infrared microscope head), and a computer. When the linear laser heat source scans the crack along the surface of the component under test, the infrared thermal imager simultaneously begins to acquire and store a sequence of thermal response characteristic images. The computer collaboratively controls the thermal excitation device and the thermal infrared acquisition device, including controlling the laser path, laser parameters, thermal imager frame rate, export of thermal image temperature data, and device start / stop.

[0084] like Figure 5 As shown, an infrared camera was used to acquire a sequence of dynamic thermal images containing multiple frames. Figure 5 As shown in (a)), the background image is subtracted frame by frame. Figure 5 (as shown in (b)); input the time step to calculate the time derivative, the time step corresponding to the frame rate of the thermal imager; for two adjacent thermal images, calculate the time gradient of temperature, and use the optical flow method to solve for the time gradient along the time gradient. Axial direction and The velocity vector of heat flow in the axial direction and Furthermore, row and column gradients are calculated for each frame of thermal image; post-processing methods using infrared thermal imaging spatiotemporal information fusion are employed to obtain calculation results for a set of two adjacent thermal images. Figure 5 As shown in (c)); the calculation results for each two adjacent thermal images are accumulated ( Figure 5 As shown in (d) in the figure, a full-field thermal response image with surface crack defects was obtained. Figure 5 (e) in the middle.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A non-destructive testing method based on spatiotemporal information fusion of infrared thermal imaging, characterized in that, include: A fast-moving linear laser is used as a thermal excitation source to scan and excite the surface of a metal specimen. A dynamic infrared thermal image sequence containing defect damage information is collected in the detection area using a thermal infrared acquisition device. The dynamic infrared thermal image sequence is processed by a post-processing method that integrates spatiotemporal information to obtain a full-field thermal response image that highlights surface crack defects. The post-processing method for the fusion of spatiotemporal information from infrared thermal imaging includes: Each frame of the dynamic infrared thermal image sequence is M×N in size, and its matrix expression is: (4) In the formula, T represents the acquired raw infrared thermal image, i represents the frame index in the dynamic thermal sequence, and L... t This indicates the number of frames contained in the dynamic thermal sequence, and 'a' represents an element in the original infrared thermal image. Calculate the spatial temperature gradient for all pixels in each frame of the thermal image; The row gradient is obtained by calculating the temperature gradient of each row in a single frame of thermal image, using the following formula: (5) In the formula, k and j represent the row index and column index of each frame of thermal image; and These represent the temperature values ​​in the j-th row and (k+1)-th column of the i-th frame of the thermal image, respectively; The matrix expression of equation (5) is: (6) In the formula, Represents the elements in the row gradient matrix; The column gradient is obtained by calculating the temperature gradient of each column in a single frame of thermal image, using the following formula: (7) In the formula, and These represent the temperature values ​​in the k-th column and (j+1)-th row of the i-th frame of the thermal image, respectively; The matrix expression of equation (7) is: (8) In the formula, Represents the elements in the column gradient matrix; Calculate the time-temperature gradient for all pixels in two adjacent thermal images: (9) In the formula, This represents the sampling time interval of the thermal imager, which is inversely proportional to the thermal imager's acquisition frame rate. The matrix expression of equation (9) is: (10) In the formula, Represents the elements in the time-temperature gradient matrix; Calculate the path along the edge using the optical flow method for two adjacent thermal images. Axial direction and The velocity vector of heat flow in the axial direction is given by the following formula: (11) In the formula, and These represent the horizontal components of the heat flow velocity. and vertical components α represents the regularization parameter, used to balance data terms and smoothing terms; β is the number of iterations in the algorithm. The matrix expression of equation (11) is: (12) In the formula, and These represent the elements in the velocity vector matrix of heat flow along the x-axis and y-axis, respectively; The formula for calculating the final defect detection result by fusing temporal and spatial dimensional information is as follows: (13) The matrix expression of equation (13) is: (14)。 2. The non-destructive testing method for infrared thermal imaging spatiotemporal information fusion according to claim 1, characterized in that, The obtained full-field thermal response image, which highlights surface crack defects, includes: The background image without thermal excitation source is subtracted frame by frame from the dynamic thermal image sequence containing multiple frames to obtain a dynamic thermal image sequence containing multiple frames with background noise removed. The data is then processed using the infrared thermal imaging spatiotemporal information fusion post-processing method established in this invention. The input time step corresponds to the frame rate of the thermal imager to calculate the time derivative. The optical flow method is used to solve for the velocity vectors of heat flow along the x-axis and y-axis. and This yields a full-field thermal response image that highlights information about surface crack defects.

3. The non-destructive testing method for infrared thermal imaging spatiotemporal information fusion according to claim 2, characterized in that, The optical flow method is used to solve for the velocity vectors of heat flow along the x-axis and y-axis. and include: Assuming constant brightness, a feature parameter pixel value Within time dt, the pixel values ​​of the feature parameters move a distance of (dx, dy) to the next frame. The spatial changes of these pixel values ​​are continuous, and the temperature distribution does not change drastically. The formula is as follows: (1) In the formula, Let dx be the temperature value of the feature parameter pixel (x, y) at time t, dy be the displacement in the x direction, dt be the displacement in the y direction, and dt be the time increment.

4. The non-destructive testing method for infrared thermal imaging spatiotemporal information fusion according to claim 3, characterized in that, Based on the assumptions of small motion and displacement, continuity, and differentiability, the propagation and diffusion of heat between adjacent frames is continuous, smooth, and differentiable. A Taylor expansion of equation (1) yields the following formula: (2) In the formula, Let represent a second-order infinitesimal; combining equations (1) and (2), we can obtain: (3) Based on the locality assumption, local thermal diffusion characteristics are captured by calculating the heat flow velocity vector of each pixel or small region. By minimizing a global energy function that includes data terms and a smoothing term to add additional constraints, the Horn-Schunck algorithm is used to solve for the energy along the path. Axial direction and The velocity vector of heat flow in the axial direction and .

Images