A method for detecting surface cracks of a cement pole based on laser technology

Abstract

The application discloses a kind of based on laser technology's cement pole surface crack detection method, it is related to nondestructive testing technical field, the method includes: deployment detection device, laser thermal excitation is carried out to the surface of pole and high frame rate infrared thermal image sequence is synchronously collected;Resolution enhancement is carried out to image, adaptive bilateral filtering and contrast enhancement;Constructing space-time gradient feature map to highlight crack thermal anomaly;Optimization identification and fitting crack path;In combination with thermal response delay time, crack depth is reversed, and the structured report containing size, positioning and risk level is output.The present application aims to solve the problem that traditional detection means is weak in rough concrete surface signal, poor in anti-interference ability, difficult to identify microcrack, realizes high sensitivity, high stability detection to the width of 0.05 millimeter microcrack, and has environmental disturbance compensation and cloud AI review ability, significantly improves the automation, standardization and engineering practicability of cement pole crack detection.

Description

Technical Field

[0001] This invention belongs to the field of nondestructive testing technology, specifically relating to a method for detecting surface cracks in cement poles based on laser technology. Background Technology

[0002] With the continuous expansion of power infrastructure, cement poles, as core supporting components of overhead transmission lines, are directly related to the safety and stability of power grid operation due to their structural integrity. During long-term service, cement poles are prone to developing micron- to millimeter-sized cracks on their surfaces due to factors such as temperature cycling, humidity erosion, mechanical loads, and material aging. Although these cracks are initially difficult to detect with the naked eye, they significantly weaken the pole's bending strength and durability, thus inducing a risk of fracture. Therefore, there is an urgent need for a non-contact surface crack detection method suitable for field use, with high sensitivity and strong environmental adaptability, to achieve rapid assessment and preventative maintenance of the health status of cement poles. Laser technology, with its advantages of non-invasiveness, high spatial resolution, and remote operation, has been widely used in non-destructive testing of metals and composite materials. However, it still faces fundamental adaptation challenges in crack identification of rough, heterogeneous, and low-reflectivity concrete structures.

[0003] Laser-driven crack detection methods primarily rely on the physical response signals generated by the interaction between laser and material, such as changes in thermal radiation, acoustic emission, or scattered light intensity. Surface or near-surface defect information is then retrieved by analyzing these response characteristics. The core of this approach lies in establishing a mapping relationship between laser parameters, material properties, and crack geometry, and capturing weak anomalous signals using a high-precision sensing system. However, existing laser detection systems are mostly designed for industrial objects with high reflectivity and smooth surfaces. Their signal models and processing algorithms are difficult to transfer to real-world working conditions such as cement poles, which exhibit strong scattering, low thermal conductivity, curved geometry, and complex surface textures, including weathered layers, stains, and moss deposits.

[0004] In existing technologies, the concrete bridge crack movement detection device with publication number CN104267043B, although employing laser-assisted positioning, is essentially a visible light image recognition system. It does not utilize laser-induced physical field changes for crack identification, resulting in a severely insufficient ability to detect early micro-cracks. Meanwhile, the dot matrix laser 3D reconstruction scheme with publication number CN108287164B heavily relies on stable laser echo signals, easily generating severe speckle noise and missing point clouds on the rough surface of cement poles, leading to distorted 3D morphology and blurred crack edges. Furthermore, neither scheme considers interference factors such as strong outdoor sunlight, wind disturbance, and vibration from high-altitude work platforms, resulting in poor system robustness. Additionally, the equipment is bulky and deployment relies on complex mechanical structures, failing to meet the needs of single-person portability and rapid on-site inspection. More importantly, existing methods lack optimized design for the correlation model between concrete thermal properties and crack depth-thermal response time, resulting in the thermal anomaly signal in the microcrack region being submerged by background noise under millisecond-level laser pulse excitation, with the signal-to-noise ratio below the effective identification threshold. Summary of the Invention

[0005] The purpose of this invention is to provide a method for detecting surface cracks in cement poles based on laser technology, so as to solve the problems of weak signal, poor anti-interference ability, and difficulty in identifying microcracks on rough concrete surfaces in existing technologies.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] A method for detecting surface cracks in cement poles based on laser technology includes the following specific steps:

[0008] Step 1: Deploy a laser thermal excitation and infrared dynamic imaging collaborative detection system, integrating a pulsed laser, an infrared thermal imager, an inertial stabilization platform, and an ambient light compensation module into a lightweight portable detection device. The device is fixed to the top of a telescopic support rod via a three-axis gimbal, enabling non-contact scanning coverage of the surface of cement poles of different heights. The laser emits a near-infrared pulsed beam with a wavelength of 1064 nanometers, a pulse width of 8 milliseconds, and an energy density of 1.2 joules per square centimeter, which is vertically incident on the target area on the pole surface, exciting a local transient temperature rise.

[0009] Step 2: Acquire an infrared dynamic thermal image sequence. At the same time as the laser pulse excitation begins, start a high-speed infrared thermal imager to continuously capture the temperature field evolution process of the target area at a sampling frequency of 120 frames per second. Record the complete thermal response time series data for a duration of 2 seconds to form an original thermal image sequence consisting of 240 single-channel grayscale images. Each frame has a resolution of 640 x 512 pixels and a temperature sensitivity better than 0.05 degrees Celsius.

[0010] Step 3: Perform multi-stage image preprocessing. First, the original thermal image sequence is enhanced in spatial resolution based on bicubic interpolation and upsampled to 1280 x 1024 pixels. Then, an adaptive bilateral filtering algorithm is used to suppress high-frequency speckle noise and background thermal drift interference, while preserving the edge thermal anomaly features corresponding to the microcrack area. Finally, histogram equalization is used to improve the overall contrast and output the enhanced thermal image sequence.

[0011] Step 4: Construct a spatiotemporal gradient feature map. For each pixel in the enhanced thermal image sequence, calculate its temperature change rate over time, and combine this with the temperature gradient magnitude in the spatial neighborhood to generate a two-dimensional spatiotemporal response intensity map, where the response intensity... The formula is as follows:

[0012]

[0013] Where α and β are the weighting coefficients of the time derivative and spatial gradient, respectively, set to 0.7 and 0.3, T is temperature, and t is time. This represents the temperature gradient amplitude.

[0014] Step 5: Identify potential crack regions. Use the Otsu adaptive threshold segmentation algorithm to binarize the spatiotemporal response intensity map and extract high response intensity connected regions as candidate crack regions. Then, apply morphological closing operation to eliminate small fracture gaps and exclude artifact regions with an area <16 pixels or an aspect ratio <3 by using minimum bounding rectangle constraint.

[0015] Step 6: Verify the geometric consistency of the crack. Perform sub-pixel level edge fitting on the retained candidate regions, use an improved Snake model for curve optimization, initialize the contour line to approximate the region boundary, and use the energy function to find the optimal crack centerline path until a smooth crack centerline path that conforms to the physical laws of heat conduction is obtained. The iterative solution of the energy function E is shown below:

[0016]

[0017] in To ensure data fidelity, weights are used to control sensitivity to thermal feature gradients. To smooth out the regularization weights, the intensity of the path curvature penalty is controlled. The curve represents the profile curvature, S is the arc length parameter, I is the position coordinate along the path, and I is the response intensity. For the response intensity gradient;

[0018] Step 7: Quantify crack parameters and output an inspection report. Measure the length, maximum width, estimated average depth, and orientation angle of each crack. Finally, generate a structured inspection report containing coordinate location, dimensional data, and risk level annotations. The depth d is obtained by inversion using the following empirical formula:

[0019]

[0020] in The crack depth. This is the peak thermal response delay time at this location. The thermal diffusivity of concrete is given by [value]. square meters per second The material calibration factor is set to 1.4.

[0021] Preferably, in step 1, the pulsed laser is an Nd:YAG solid-state laser source, which has high peak power and narrow pulse width characteristics, ensuring a significant thermal excitation effect under low energy deposition conditions and avoiding ablation damage to the concrete surface. At the same time, the laser spot diameter is adjusted to 30 mm to form a uniform circular heating zone, covering the typical microcrack propagation scale range.

[0022] Preferably, in step 2, the infrared thermal imager is equipped with a cooled indium antimonide detector with a working wavelength of 3.7 to 4.8 micrometers, which matches the peak wavelength of concrete radiation, improves the signal-to-noise ratio, and has a built-in non-uniformity correction module that automatically performs blackbody calibration once per minute to eliminate sensor drift caused by long-term operation.

[0023] Preferably, in step 3, the standard deviation of the spatial domain kernel function of the adaptive bilateral filter... Set to 1.5 pixels, grayscale kernel function standard deviation The temperature is dynamically adjusted based on the local temperature variance, ranging from 0.1 to 0.4 degrees Celsius, thereby smoothing out noise while maintaining the sharpness of the crack heat shadow boundary.

[0024] Preferably, the weighting coefficients in step 4 and Adjustments are made in real time based on on-site lighting conditions and wind speed. When the ambient light intensity exceeds 50,000 lux or the wind speed is greater than 5 meters per second, the adjustment is increased. The value was reduced to 0.85 to highlight the dynamic characteristics in the time domain and reduce the impact of spatial noise.

[0025] Preferably, in step 5, the Otsu algorithm calculates the global optimal threshold T to maximize the variance between the foreground and background classes. After initial binarization, connected component analysis is introduced to retain only areas with an area between 16 and 2048 pixels, and targets with a circularity > 0.3 are further removed to prevent misjudgment of irregular attachments such as water stains and moss.

[0026] Preferably, in step 6, the energy function of the Snake model introduces an external guiding force term, the direction of which points to the maximum value of the spatiotemporal gradient, guiding the contour to actively migrate towards the edge of the real crack. At the same time, the maximum number of iterations is set to 50, and the convergence tolerance is 0.01 pixels to ensure a balance between computational efficiency and accuracy.

[0027] Preferably, in step 7, the crack orientation angle is measured clockwise from due north to the main crack axis to assess the correlation between load and stress direction. The risk level is determined based on the combination of length and depth. Cracks with a length > 200 mm and an estimated depth > 8 mm are marked as Level 3 high-risk defects and must be immediately discontinued and replaced.

[0028] Preferably, it also includes an environmental disturbance compensation mechanism, which simultaneously collects data on ambient temperature, relative humidity, wind speed, and solar radiation angle during the detection process, and constructs a multivariate regression model to correct the thermal response delay time. It compensates for the deep inversion bias caused by fluctuations in external conditions and improves the consistency of detection under different climate conditions.

[0029] Preferably, it also includes remote data transmission and cloud diagnostic functions. The detection device has a built-in 4G communication module to encrypt and upload the original thermal image sequence, intermediate processing results and final report to the central server. The background deploys a convolutional neural network model to perform secondary verification of suspicious areas. The network structure includes 12 convolutional layers and 3 fully connected layers. The activation function is LeakyReLU, and the classification accuracy reaches 97.6% in actual tests.

[0030] Preferably, the method is applicable to various prestressed concrete poles with diameters ranging from 300 to 600 mm and heights ranging from 8 to 15 meters. The inspection operation can be completed within a distance of 1.5 to 3 meters from the pole surface, with a single scan area of ​​0.7 square meters. The entire process takes no more than 90 seconds and supports continuous inspection of no less than 50 poles.

[0031] Compared with the prior art, the beneficial technical effects of the present invention are as follows:

[0032] This invention fundamentally solves the core problems of weak signal and poor anti-interference ability of traditional detection methods on rough, low-reflectivity concrete surfaces by constructing a technical path that deeply integrates pulsed laser thermal excitation and infrared dynamic imaging. The millisecond-level controllable laser heating method can effectively excite the unique heat conduction blocking effect in the microcrack area without damaging the material, forming identifiable local temperature difference anomalies. Combined with a high frame rate infrared imaging system, it can completely capture the entire process of thermal field evolution, breaking through the limitation of static imaging in losing transient features.

[0033] This invention establishes a spatiotemporal gradient joint feature extraction model, which enables the separation of weak crack signals from complex backgrounds, significantly improving the signal-to-noise ratio and detection sensitivity, and can identify early hairline cracks with a width as low as 0.05 mm.

[0034] This invention uses adaptive image enhancement and Snake model-driven edge optimization strategies to ensure stable and reliable detection results even in field conditions with wind disturbance, vibration, and ambient light interference.

[0035] This invention, by integrating into a portable platform, eliminates the reliance on large mechanical support structures and meets the on-site requirements of single-person operation and rapid deployment; by fusing prior knowledge of material thermophysical properties with crack parameter inversion algorithms, it improves the engineering practicality of depth estimation; and through the supporting remote transmission and AI-assisted diagnostic functions, it provides an intelligent data management foundation for large-scale inspection of power facilities, and fully realizes the standardization, automation and informatization upgrade of the inspection process. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the overall technical solution architecture of the laser technology-based method for detecting surface cracks in cement poles proposed in this invention.

[0037] Figure 2 This is a flowchart illustrating the core principle of the collaborative detection of pulsed laser thermal excitation and infrared dynamic imaging in this invention. Detailed Implementation

[0038] The features and exemplary embodiments of various aspects of the present invention will now be described in detail. To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely intended to explain the present invention and not to limit the present invention. For those skilled in the art, the present invention can be practiced without some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present invention by illustrating examples of the invention.

[0039] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0040] In the embodiments of the present invention, the same reference numerals denote the same components, and for the sake of brevity, detailed descriptions of the same components are omitted in different embodiments. It should be understood that the thickness, length, width, and other dimensions of various components in the embodiments of the present invention shown in the accompanying drawings, as well as the overall thickness, length, width, and other dimensions of the integrated device, are merely illustrative and should not constitute any limitation on the present invention; the term "multiple" in the present invention refers to two or more (including two).

[0041] Example 1

[0042] Currently, with the continuous expansion of power infrastructure, cement poles, as core supporting components of overhead transmission lines, directly affect the safety and stability of the power grid operation due to their structural integrity. During long-term service, cement poles are prone to developing micron- to millimeter-sized cracks on their surfaces due to factors such as temperature cycling, humidity corrosion, mechanical loads, and material aging. Although these cracks are initially difficult to detect with the naked eye, they significantly weaken the pole's bending strength and durability, thus inducing a risk of fracture. Therefore, there is an urgent need for a non-contact surface crack detection method suitable for field use, possessing high sensitivity and strong environmental adaptability, to achieve rapid assessment and preventative maintenance of the health status of cement poles. Laser technology, with its advantages of non-invasiveness, high spatial resolution, and remote operation, has been widely used in non-destructive testing of metals and composite materials, but it still faces fundamental adaptation challenges in crack identification for rough, heterogeneous, and low-reflectivity concrete structures. To address the aforementioned technical problems, this invention proposes a technical approach that deeply integrates pulsed laser thermal excitation with infrared dynamic imaging. This approach fundamentally solves the core challenges of weak signals and poor anti-interference capabilities of traditional detection methods on rough, low-reflectivity concrete surfaces. The invention is then applied to a laser-based method for detecting surface cracks in cement poles.

[0043] Reference Appendix Figure 1The overall technical architecture of this invention includes a pulsed laser, an infrared thermal imager, an inertial stabilization platform, and an ambient light compensation module deployed in a lightweight portable detection device. The device is fixed to the top of a retractable support rod via a three-axis gimbal, enabling non-contact scanning coverage of the surface of cement poles at different heights. (See attached document.) Figure 2 Its core principle is to use millisecond-level near-infrared pulse lasers to be vertically incident on the target area of ​​the cement pole surface to excite local transient temperature rise; the microcrack area exhibits different thermal response characteristics from the intact concrete area due to the obstruction of the heat conduction path. This difference is captured by a high-speed infrared thermal imager and converted into a dynamic thermal image sequence. After multi-stage image processing and feature extraction, the automatic identification, geometric verification and parameter quantification of cracks are finally realized.

[0044] In the aforementioned laser-based method for detecting surface cracks on cement poles, step 1 involves deploying a laser thermal excitation and infrared dynamic imaging collaborative detection system. This system integrates a pulsed laser, an infrared thermal imager, an inertial stabilization platform, and an ambient light compensation module into a lightweight, portable detection device. The device is fixed to the top of a retractable support pole via a three-axis gimbal, enabling non-contact scanning coverage of cement pole surfaces of varying heights. The laser emits a near-infrared pulsed beam with a wavelength of 1064 nm, a pulse width of 8 ms, and an energy density of 1.2 joules per square centimeter, which is vertically incident on the target area of ​​the pole surface, triggering a localized transient temperature rise. Specifically, the pulsed laser uses an Nd:YAG solid-state laser source with a peak power of up to 500 watts, and the pulse repetition frequency is set to 1 Hz to ensure concentrated and controllable energy for each excitation. The laser spot diameter is adjusted to 30 mm using a combination of a beam expander and a focusing lens, forming a uniform circular heating area covering the typical microcrack propagation scale (0.05 to 5 mm width). Experimentally verified, this energy density can generate a transient temperature rise of approximately 3 to 5 degrees Celsius on the concrete surface, sufficient to excite detectable thermal anomaly signals while avoiding false positive responses caused by surface ablation or moisture evaporation due to overheating. The inertial stabilization platform incorporates a three-axis gyroscope and accelerometer with a sampling frequency of 1 kHz, sensing real-time changes in device attitude. It also uses a piezoelectric ceramic actuator to provide micro-arc-level active compensation for the infrared thermal imager and laser output port, effectively suppressing image blurring caused by vibration and wind disturbance of the aerial work platform. The ambient light compensation module consists of a visible light shield and an active LED array. Under strong sunlight, the supplementary lighting is turned off; on cloudy days or in backlit areas, the 470nm blue LED automatically activates to assist in device positioning and background texture reference, but does not participate in thermal signal acquisition.

[0045] In the aforementioned laser-based method for detecting surface cracks in cement poles, step 2 involves acquiring a sequence of infrared dynamic thermal images. Simultaneously with the laser pulse excitation, a high-speed infrared thermal imager is activated to continuously capture the temperature field evolution of the target area at a sampling frequency of 120 frames per second. Complete thermal response time-series data is recorded for a duration of 2 seconds, forming an original thermal image sequence consisting of 240 single-channel grayscale images. Each frame has a resolution of 640 x 512 pixels, and the temperature sensitivity is better than 0.05 degrees Celsius. Specifically, the infrared thermal imager is equipped with a cooled indium antimonide detector, operating in the 3.7 to 4.8 micrometer band. This band highly matches the peak wavelength of blackbody radiation (approximately 4.2 micrometers) of concrete at room temperature, thereby maximizing the signal-to-noise ratio. The detector is cooled to 77 Kelvin to effectively suppress dark current noise. The infrared thermal imager incorporates a non-uniformity correction module, automatically performing blackbody calibration every minute. The device integrates a miniature blackbody cavity with a constant temperature of 30 degrees Celsius. During calibration, the shutter briefly closes to acquire a uniform radiation field across the entire image, updating the gain and bias parameters of each pixel and eliminating sensor drift caused by prolonged operation. The image acquisition trigger signal is synchronously output from the laser's Q-switch, ensuring precise zero-point alignment. The raw thermal image sequence is stored in a 16-bit lossless format on a solid-state drive, with each frame containing a microsecond-level accurate timestamp and global temperature offset, providing fundamental data for subsequent spatiotemporal analysis.

[0046] In the aforementioned laser-based method for detecting surface cracks on cement poles, step 3 involves multi-stage image preprocessing. First, the original thermal image sequence undergoes spatial resolution enhancement based on bicubic interpolation, upsampling it to 1280 x 1024 pixels. Then, an adaptive bilateral filtering algorithm is used to suppress high-frequency speckle noise and background thermal drift interference, preserving the edge thermal anomaly features corresponding to the microcrack region. Finally, histogram equalization is used to improve overall contrast, outputting the enhanced thermal image sequence. Specifically, the bicubic interpolation algorithm generates new pixels by calculating the weighted average of 16 neighboring pixels, with the weighting function being a cubic spline curve. This effectively avoids the block effect of nearest-neighbor interpolation and the blurring effect of bilinear interpolation, ensuring that the microcrack edges remain clear after upsampling. The standard deviation of the spatial domain kernel function of the adaptive bilateral filter is used. Set to 1.5 pixels, grayscale kernel function standard deviation Dynamic adjustment based on local temperature variance: For each frame of image, the local variance is calculated by sliding a 9x9 pixel window. ,like <0.01 degrees Celsius squared, then Use 0.4 degrees Celsius for strong smoothing; if >0.1 degrees Celsius squared, then The temperature was reduced to 0.1 degrees Celsius to preserve details; linear interpolation was used for intermediate values. This strategy effectively preserved the steep temperature gradient boundaries of the crack region caused by thermal resistance while smoothing high-frequency speckle caused by atmospheric turbulence and electronic noise. Histogram equalization employed cumulative distribution function transformation to map the original 8-bit grayscale range to the full dynamic range, significantly improving the visual recognizability of weak thermal anomaly regions and laying the foundation for subsequent feature extraction.

[0047] In the aforementioned method for detecting surface cracks in cement poles based on laser technology, step 4 involves constructing a spatiotemporal gradient feature map. For each pixel in the enhanced thermal image sequence, the rate of temperature change over time is calculated, and combined with the temperature gradient amplitude within the spatial neighborhood, a two-dimensional spatiotemporal response intensity map is generated. Defined by the formula, and The weighting coefficients for the time derivative and spatial gradient are set to 0.7 and 0.3, respectively. Specifically, the time derivative... Calculated using the central difference method, for the ... The temperature change rate of frame pixel (x, y) is as follows:

[0048]

[0049] Where Δt is 1 / 120 of a second, T is the temperature, and x and y are the spatial coordinates;

[0050] Spatial gradient The Sobel operator is used to calculate the partial derivatives in the horizontal and vertical directions respectively, and then the gradient magnitude is synthesized, as shown in the following formula:

[0051]

[0052] Where T is temperature. It is the partial derivative in the horizontal direction. It is the partial derivative in the vertical direction.

[0053] The construction of the spatiotemporal response intensity map I(x,y) incorporates two key physical characteristics of the crack region: firstly, the heat conduction at the microcrack is impeded, leading to a higher rate of temperature rise. The crack is significantly smaller than the surrounding intact area, exhibiting a negative anomaly; secondly, there is a clear spatial temperature jump at the crack edge, i.e. The values ​​are relatively high. Through weighted linear combination, this feature map can effectively amplify the crack signal and suppress the uniform background. Weighting coefficients and The system adjusts in real-time based on ambient lighting conditions and wind speed: It incorporates a built-in light sensor and ultrasonic anemometer; when ambient light intensity exceeds 50,000 lux or wind speed is >5 m / s, the adjustment is increased. The value is set to 0.85 to highlight the dynamic characteristics in the time domain and reduce the impact of spatial noise caused by strong light reflection or wind-induced convection; otherwise, the default values ​​of 0.7 and 0.3 are maintained. This adaptive mechanism significantly improves the robustness of the system in complex field environments and the response strength. The formula is as follows:

[0054]

[0055] Where α and β are the weighting coefficients of the time derivative and spatial gradient, respectively, set to 0.7 and 0.3, T is temperature, and t is time. This represents the temperature gradient magnitude.

[0056] In the aforementioned laser-based method for detecting surface cracks on cement poles, step 5 involves identifying potential crack regions. The spatiotemporal response intensity map is binarized using the Otsu adaptive threshold segmentation algorithm. High-response-intensity connected regions are extracted as candidate crack regions. Morphological closing operations are then applied to eliminate small fracture gaps, and artifact regions with an area <16 pixels or an aspect ratio <3 are excluded using minimum bounding rectangle constraints. Specifically, the Otsu algorithm iterates through all possible thresholds T, calculates the inter-class variance between the foreground (high response) and background (low response), and selects the T that maximizes the variance as the globally optimal threshold. After initial binarization, eight-connected-domain analysis is introduced, retaining only regions with areas between 16 and 2048 pixels—16 pixels correspond to the lower limit of the projected area of ​​a 0.05 mm wide, 50 mm long crack at a resolution of 1280 x 1024, while 2048 pixels cover the vast majority of engineering-related cracks. Subsequently, the minimum bounding rectangle is calculated for each connected component. If its area is less than 16 pixels or its aspect ratio is less than 3, it is identified as a noise point, dust, or random hotspot and is removed. To further prevent misjudgment of irregular attachments such as water stains and moss, the circularity of each region is calculated. Where A is the area of ​​the region and P is the perimeter of the region, targets with C > 0.3 are eliminated because real cracks are usually long and narrow lines with a circularity far below this threshold. Morphological closing operations use 5x5 pixel square structuring elements, first dilating and then eroding to effectively connect the crack fracture parts caused by noise or local signal weakening, forming continuous candidate regions.

[0057] In the above-mentioned method for detecting surface cracks on cement poles based on laser technology, step 6 involves verifying the geometric consistency of the cracks, performing sub-pixel-level edge fitting on the retained candidate regions, optimizing the curves using an improved Snake model, initializing the contour lines to approximate the region boundaries, and iteratively solving the energy function until convergence to obtain a smooth crack centerline path that conforms to the physical laws of heat conduction. The iterative solution of the energy function E is shown below:

[0058]

[0059] in To ensure data fidelity, weights are used to control sensitivity to thermal feature gradients. To smooth out the regularization weights, the intensity of the path curvature penalty is controlled. The curve represents the profile curvature, S is the arc length parameter, I is the position coordinate along the path, and I is the response intensity. For the response intensity gradient, where and As a weighting factor, Indicates the curvature of the profile.

[0060] Specifically, the energy function of the Snake model introduces an external guiding force term, the direction of which points to the spatiotemporal gradient magnitude. At its maximum, the profile is guided to actively migrate towards the edge of the actual crack. The internal energy term includes the first derivative (controlling profile tension) and the second derivative (controlling profile stiffness), while the external energy term is driven inversely by the gradient of the spatiotemporal response intensity map. During initialization, the binary boundary of the candidate region is used as the initial profile, and a discrete point set with sub-pixel precision is obtained through cubic spline interpolation. During iteration, the Euler-Lagrange equation is solved using the finite difference method to update the profile point positions. Weighting factors... Set it to 0.6. The value is set to 0.4 to balance contour smoothness and edge fit. The maximum number of iterations is set to 50, and the convergence tolerance is 0.01 pixels, meaning the iteration terminates when the root mean square displacement of the contour points in two consecutive iterations is less than 0.01 pixels. This strategy ensures that the crack centerline path is not only geometrically smooth but also strictly follows the trajectory of the strongest thermal anomaly signal, conforming to the physical laws of heat conduction and effectively eliminating nonlinear artifacts.

[0061] In the aforementioned laser-based method for detecting surface cracks on cement poles, step 7 involves quantifying crack parameters and outputting a detection report. This includes measuring the length, maximum width, estimated average depth, and orientation angle of each crack. The depth d is determined using an empirical formula. Inversion yielded, among which The crack depth. This is the peak thermal response delay time at this location. The thermal diffusivity of concrete is given by [value]. square meters per second The material calibration coefficient is set to 1.4, ultimately generating a structured inspection report containing coordinate positioning, dimensional data, and risk level annotations. Specifically, the crack length is calculated by integrating along the optimized centerline path; the maximum width is obtained by searching for the maximum distance between the two edge points in the centerline normal direction; and the average depth is estimated. Based on the theory of heat wave propagation: the greater the depth of the microcrack, the longer the time required for the heat wave to penetrate, which manifests as a longer peak delay time τ in the thermal response. In the empirical formula... The peak time was obtained by fitting a Gaussian function to the temperature-time curves at each point on the center line. Take typical values ​​of concrete square meters per second; =1.4 Determined through laboratory calibration, regression fitting was performed using artificial crack samples with known depths ranging from 0.5 to 10 mm. The crack orientation angle was measured clockwise to the principal axis of the crack (determined by principal component analysis) with due north as the reference, used to assess the correlation between load and stress directions. Risk levels were classified based on the combination of length and depth: Level 1 (low risk) was less than 50 mm in length or less than 2 mm in depth; Level 2 (medium risk) was 50 to 200 mm in length and 2 to 8 mm in depth; Level 3 (high risk) was greater than 200 mm in length and an estimated depth exceeding 8 mm, requiring immediate discontinuation and replacement. The inspection report was generated in JSON format, including GPS coordinates, pole number, crack list, risk level, and original data index, and was displayed locally on an LCD screen.

[0062] Furthermore, this invention also includes an environmental disturbance compensation mechanism, which simultaneously collects data on ambient temperature, relative humidity, wind speed, and solar radiation angle during the detection process, and constructs a multivariate regression model to correct the thermal response delay time. This compensates for depth inversion bias caused by fluctuations in external conditions, improving detection consistency across climatic conditions. Environmental sensor data is aligned with the timestamps of thermal image sequences, input into a pre-trained ridge regression model, and outputs a correction factor. Used to correct the original The model is calibrated under a variety of climatic conditions, including -10 to 45 degrees Celsius, 30% to 90% relative humidity, and 0 to 10 m / s wind speed, ensuring that the depth estimation error is controlled within ±1 mm.

[0063] Furthermore, this invention also includes remote data transmission and cloud-based diagnostic functions. The detection device has a built-in 4G communication module that encrypts and uploads the original thermal image sequence, intermediate processing results, and final report to the central server. A convolutional neural network model is deployed in the background to perform secondary verification on suspicious areas. The network structure includes 12 convolutional layers and 3 fully connected layers, and the activation function is LeakyReLU. The measured classification accuracy reaches 97.6%. This neural network is trained on a dataset containing 100,000 labeled samples, including real cracks, water stains, moss, shadows, etc. It can effectively identify complex crack shapes that the Snake model may miss or eliminate stubborn artifacts, forming a dual protection mechanism of human-machine collaboration.

[0064] To verify the effectiveness of this invention, the following specific application example was constructed: During an inspection of a 10 kV distribution line in North China, the concrete pole to be inspected was a tapered prestressed concrete pole, 450 mm in diameter, 12 meters high, and with a service life of 8 years. The inspector held the device of this invention, extending the telescopic support rod to 2.5 meters, and fine-tuned it using a three-axis gimbal to align the laser spot center with a suspected area 2 meters above the ground on the pole. The device automatically collected environmental parameters: air temperature 25 degrees Celsius, humidity 60%, wind speed 2 meters per second, and illumination 40,000 lux, triggering laser pulses and infrared acquisition. After 2 seconds, the system completed the entire processing flow, highlighting a longitudinal crack on the screen with a length of 185 mm, a maximum width of 0.12 mm, an estimated average depth of 6.3 mm, and a directional angle of 175 degrees, classifying it as a level 2 medium-risk defect. Technicians then marked the pole and included it in the monthly tracking and monitoring plan. The entire process took 82 seconds, and the remaining battery power supported the inspection of 48 more poles. This example fully demonstrates the efficiency, accuracy, and engineering applicability of the invention in real-world field scenarios.

[0065] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape, and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for detecting surface cracks in cement poles based on laser technology, characterized in that, Includes the following steps: Step 1: Deploy a laser thermal excitation and infrared dynamic imaging collaborative detection system, integrating a pulsed laser, an infrared thermal imager, an inertial stabilization platform, and an ambient light compensation module into a lightweight portable detection device; Step 2: Acquire a sequence of infrared dynamic thermal images. At the same time as the laser pulse excitation begins, start the high-speed infrared thermal imager to capture the temperature field evolution process of the target area, record the complete thermal response time series data, and form the original thermal image sequence. Step 3: Perform multi-stage image preprocessing. First, perform resolution enhancement, adaptive bilateral filtering, and contrast enhancement on the original thermal image sequence, and output the enhanced thermal image sequence. Step 4: Construct a spatiotemporal gradient feature map. For each pixel in the enhanced thermal image sequence, calculate its temperature change rate in the time dimension, and combine it with the temperature gradient magnitude in the spatial neighborhood to generate a two-dimensional spatiotemporal response intensity map. Step 5: Identify potential crack regions. Use the Otsu adaptive threshold segmentation algorithm to binarize the spatiotemporal response intensity map and extract high response intensity connected regions as candidate crack regions. Then, apply morphological closing operation to eliminate small fracture gaps and eliminate artifact regions with an area <16 pixels and an aspect ratio <3 by using minimum bounding rectangle constraint. Step 6: Verify the geometric consistency of the crack. Perform sub-pixel level edge fitting on the retained candidate regions. Use the improved Snake model for curve optimization. Initialize the contour line to approximate the region boundary. Iterate multiple times until convergence to obtain a smooth crack centerline path that conforms to the physical laws of heat conduction. Step 7: Quantify crack parameters and output inspection report. Measure the length, maximum width, average depth estimate and orientation angle of each crack, and finally generate a structured inspection report containing coordinate positioning, size data and risk level labeling.

2. The method for detecting surface cracks in cement poles based on laser technology according to claim 1, characterized in that: In step 1, the pulsed laser uses an Nd:YAG solid-state laser source, and the laser spot diameter is adjusted to 30 mm to form a uniform circular heating area. The portable detection device is fixed to the top of the telescopic support rod via a three-axis gimbal to perform non-contact scanning coverage of the surface of cement poles of different heights. The laser emits a near-infrared pulse beam with a wavelength of 1064 nm, a pulse width of 8 ms, and an energy density of 1.2 joules / square centimeter, which is vertically incident on the target area of ​​the pole surface to excite a local transient temperature rise.

3. The method for detecting surface cracks in cement poles based on laser technology according to claim 1, characterized in that: In step 2, the infrared thermal imager is equipped with a cooled indium antimonide detector, operating in the 3.7 to 4.8 micrometer band, and has a built-in non-uniformity correction module that automatically performs blackbody calibration once per minute; the target to be captured should be continuously captured at a sampling frequency of 120 frames per second for a duration of 2 seconds; the thermal response timing data should be a complete 2-second data set. The original thermal image sequence consists of 240 single-channel grayscale images, each with a resolution of 640 x 512 pixels and a temperature sensitivity better than 0.05 degrees Celsius.

4. The method for detecting surface cracks in cement poles based on laser technology according to claim 1, characterized in that: In step 3, the resolution enhancement is based on bicubic interpolation in the space, upsampled to 1280 x 1024 pixels; the standard deviation of the spatial domain kernel function of the adaptive bilateral filter... Set to 1.5 pixels, grayscale kernel function standard deviation The temperature is dynamically adjusted based on the local temperature variance, ranging from 0.1 to 0.4 degrees Celsius. The adaptive bilateral filtering algorithm aims to suppress high-frequency speckle noise and background thermal drift interference, while preserving the edge thermal anomaly characteristics corresponding to the microcrack region. The contrast enhancement is achieved through histogram equalization.

5. The method for detecting surface cracks in cement poles based on laser technology according to claim 1, characterized in that: In step 5, after the initial binarization by the Otsu algorithm, connected component analysis is introduced to retain only the area between 16 and 2048 pixels, and targets with a circularity greater than 0.3 are further removed.

6. The method for detecting surface cracks in cement poles based on laser technology according to claim 1, characterized in that: In step 6, the energy function of the Snake model introduces an external guiding force term, which points to the maximum value of the spatiotemporal gradient. The maximum number of iterations is 50, and the convergence tolerance is 0.01 pixels.

7. The method for detecting surface cracks in cement poles based on laser technology according to claim 1, characterized in that: In step 7, the crack orientation angle is measured clockwise from due north to the main crack axis. The risk level is determined based on the combination of length and depth. Cracks with a length > 200 mm and an estimated depth > 8 mm are marked as level 3 high-risk defects.

8. The method for detecting surface cracks in cement poles based on laser technology according to claim 1, characterized in that: It also includes an environmental disturbance compensation mechanism, which simultaneously collects data on ambient temperature, relative humidity, wind speed and solar radiation angle during the detection process to construct a multivariate regression model.

9. The method for detecting surface cracks in cement poles based on laser technology according to claim 1, characterized in that: It also includes remote data transmission and cloud diagnostic functions. The detection device has a built-in 4G communication module to encrypt and upload the original thermal image sequence, intermediate processing results and final report to the central server. The background deploys a convolutional neural network model to perform secondary verification of suspicious areas. The network structure includes 12 convolutional layers and 3 fully connected layers, and the activation function is LeakyReLU.

Images