Rapid evaluation method of water penetration depth and water content distribution of cement-based materials based on hyperspectral imaging technology

Abstract

The application relates to a rapid evaluation method for the water penetration depth and water content distribution of cement-based materials based on hyperspectral imaging technology, and aims to solve the problems of slow evaluation speed and low efficiency of the water penetration depth and water content distribution of existing cement-based materials. The evaluation method comprises the following steps: 1, carrying out a water absorption test experiment on the cement-based material; 2, obtaining a hyperspectral image through infrared hyperspectral camera shooting; 3, correcting the hyperspectral image; 4, obtaining a reflectivity gradient graph of the cement-based material sample along the height direction through the corrected image, and drawing a reflectivity time-varying characteristic graph along the height direction of the sample; 5, generating and mapping the corrected sample hyperspectral image by using the additive average analysis; 6, quantifying the water content and the spectral characteristics and calibrating the equation; and 7, visualizing the water penetration process and the water content. The application utilizes near-infrared hyperspectral imaging technology to scan the cement-based material, evaluates the water penetration depth, and enhances the visualization effect of the water distribution based on the additive average method.

Description

Technical Field

[0001] The present invention belongs to the technical field of cement-based material permeability testing, and in particular relates to a method for evaluating the water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology. Background Art

[0002] The water permeation behavior of cement-based materials is closely related to the service durability of structures such as buildings, water conservancy, transportation and marine engineering. On the one hand, water infiltration will produce direct physical and chemical reactions with cement-based materials, such as soft water dissolution, shrinkage and swelling, freeze-thaw damage, etc., causing the service performance of engineering materials and components to deteriorate; on the other hand, water is the main carrier for corrosive media to enter cement-based materials, such as Cl - 、SO4 2- Salt ions such as ions further accelerate the degradation of their performance and significantly reduce the service life and safety of engineering structures. Therefore, to accurately understand and evaluate the service status and durability of cement-based materials, it is necessary to first quantify the water penetration behavior of cement-based materials. The most important prerequisite is to quickly and effectively evaluate the distribution and content of water penetration in cement-based materials.

[0003] At present, the methods for evaluating the moisture penetration distribution and content of cement-based materials mainly adopt mass weighing method, nuclear magnetic resonance method, neutron imaging method and X-ray computed tomography method. However, the above methods still face huge challenges in terms of observation efficiency, test cost and application scenarios. The mass weighing method (such as invention patent ZL201310441396.9) can only evaluate moisture penetration from a macroscopic perspective by the mass change before and after water penetration. It cannot obtain the moisture distribution on the microscopic scale and the test efficiency is not high. The accuracy of the result is greatly affected by the experimental personnel's operation ability; the nuclear magnetic resonance method (such as invention patents ZL201610625326.2 and ZL202410195881.0) uses the nuclear magnetic resonance phenomenon of hydrogen atoms in water to measure moisture information, but this method is easily affected by paramagnetic ions, has high requirements on the internal composition of the cement-based material sample being tested, and has limited applicability; the neutron imaging method (such as invention patent ZL202111409416.5) realizes moisture visualization by measuring the signal attenuation of the neutron beam. Although It can accurately capture the distribution state of moisture, but this method still cannot effectively distinguish between the bound water of hydrated calcium silicate in cement-based materials and the free water infiltrating from the external environment. The experimental process is complicated and costly, and its promotion and application are greatly limited. X-ray computed tomography (such as invention patents ZL201510617455.2, ZL201810244594.9 and CN202411190762.2) uses the grayscale value changes of X-ray scanning images to determine the moisture transmission behavior, thereby realizing quantitative analysis of the moisture content and distribution in cement-based materials. This method has acceptable efficiency and accuracy, but is limited by the bulky X-ray generator. It is only suitable for laboratory test scenarios and cannot perform in-situ observations of outdoor physical engineering structures. Its application scenarios are limited. Summary of the Invention

[0004] The purpose of the present invention is to solve the problems of the existing methods for evaluating the water penetration depth and water content distribution of cement-based materials being slow, inefficient, expensive, and limited to laboratories. Instead, a method for rapidly evaluating the water penetration depth and water content distribution of cement-based materials based on a near-infrared hyperspectral camera is proposed. The raw data (image) acquisition process alone can be shortened from the current minute level to millisecond level, and the computer configuration and computing power costs required for subsequent data and image processing are also greatly reduced, fully satisfying the rapid and efficient evaluation of the water penetration process of cement-based materials. At the same time, the near-infrared hyperspectral camera is small in size and light in weight, and can be combined with drone technology to realize on-site in-situ evaluation of outdoor structures, greatly expanding the application scenarios and providing an efficient and practical research method for the quantitative evaluation of the water penetration behavior of cement-based materials and the durability of engineering structures.

[0005] The method for rapidly assessing the water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology of the present invention is implemented by the following steps:

[0006] Step 1: Water absorption test:

[0007] The bottom of the cement-based material sample is immersed in water to conduct a water absorption test experiment;

[0008] Step 2: Near-infrared hyperspectral camera scanning:

[0009] An infrared hyperspectral camera was installed above the cement-based material sample. During the water absorption test, the infrared hyperspectral camera was used to capture hyperspectral images of the sample at different water absorption times.

[0010] Step 3: Hyperspectral image correction:

[0011] Performing dark correction and white correction on the sample hyperspectral image obtained in step 2 to obtain a corrected sample hyperspectral image;

[0012] Step 4: Quick assessment of reflectance spectral gradient and water penetration depth:

[0013] The reflectivity gradient map of the cement-based material sample along the height direction is obtained from the corrected sample hyperspectral image. The reflectivity gradient map uses wavelength as the horizontal axis and reflectivity as the vertical axis. The reflectivity gradient curve of the cement-based material sample at different heights is drawn. The valley wavelength is determined from the reflectivity gradient curve, and the cement-based material sample is divided into a dry area, a transition area (including the wetting front), and a completely wetting area (corresponding to the water absorption time).

[0014] The reflectivity at the valley wavelength at different heights was obtained from the reflectivity gradient map of each cement-based material sample. The sample height was used as the horizontal axis and the reflectivity at the valley wavelength as the vertical axis to obtain the sample height-reflectivity curve for the corresponding water absorption time. The inflection point (mutation turning point) from the transition zone to the dry zone is the water penetration depth. The sample height-reflectivity curve for different water absorption times during the water absorption test experiment was plotted to obtain a time-varying characteristic diagram of the reflectivity along the sample height direction.

[0015] During the water absorption test, the square root of the water absorption time was used as the horizontal axis, and the water penetration depth was used as the vertical axis. A water penetration evaluation model inside the cement-based material was fitted to complete the water penetration depth evaluation.

[0016] Step 5: Reflectance spectrum mapping and visualization:

[0017] The calibrated hyperspectral image of the sample was generated and mapped using the added average analysis, and the area of ​​the cement-based material sample was selected as the area-of-interest (ROI).

[0018] The corrected sample hyperspectral image is imported into the Python function library in the form of a single-band gray image and converted into a two-dimensional array format, the reflectance maximum value and the reflectance minimum value are obtained respectively, and then the two-dimensional coordinates of the reflectance maximum value and the reflectance minimum value are located, the extreme values (reflectance maximum value, reflectance minimum value) are extracted as the normalized baseline and mapped on the corrected sample image, and the mapped hyperspectral image is obtained;

[0019] Step six: quantification of moisture content and spectral characteristics and calibration equation:

[0020] During the water absorption test, the cement-based material sample is cut into two halves along the height direction, and one half of the sample is cut into a thin sample along the height direction, the unit volume moisture content of the thin sample is calculated, the data points are established from the reflectance value at the valley wavelength and the unit volume moisture content, and the relationship curve between the sample valley wavelength reflectance value and the unit volume moisture content is fitted;

[0021] Step seven: moisture penetration process and moisture content visualization output:

[0022] The mapped hyperspectral image of the cement-based material sample is output, the moisture content of different regions in the mapped hyperspectral image is calibrated through the relationship curve between the sample valley wavelength reflectance value and the unit volume moisture content, and the rapid evaluation of the water content distribution of the cement-based material is completed.

[0023] The present application is aimed at the problems of slow speed, low efficiency, high cost and limited application scene of the existing evaluation method of the moisture penetration depth and water content distribution of cement-based materials. Based on the specific absorption characteristics of water molecules in the near-infrared waveband, the near-infrared hyperspectral imaging technology is used to scan the cement-based material in the water absorption process, the spectral image effect is improved by correcting the scanned image, and the reflectance gradient of the cement-based material in the water absorption process is obtained based on the reflectance difference of the image surface. Then, the water absorption process of the cement-based material is visualized based on the additive average method, and the spectral reflectance of the image is calibrated combined with the slice weighing method, realizing the real-time tracking of the moisture distribution and penetration depth evolution of the cement-based material in the water absorption process.

[0024] The innovation of the present application is that the rapid evaluation method of the moisture penetration depth and water content distribution of cement-based materials based on the hyperspectral imaging technology is:

[0025] 1) A rapid evaluation method of the moisture penetration depth of cement-based materials is proposed based on the specific absorption characteristics of water molecules in the near-infrared spectral waveband

[0026] Water molecules have specific absorption characteristics in the near-infrared band, among which the absorption peak at 1400-1450nm is dominated by the absorption of the OH group of water. This absorption phenomenon is caused by the multiplication and combination effects of the vibration modes of functional groups and is highly sensitive to moisture distribution. Utilizing this specific absorption characteristic, the spectral data of cement-based materials are acquired through hyperspectral scanning, and a reflectivity gradient along the height direction is established. Taking the peak wavelength of the absorption peak as the key parameter, an efficient and rapid evaluation of the water penetration depth inside the cement-based material is achieved. The present invention proposes a method for rapidly and quantitatively evaluating the water penetration depth of cement-based materials under complex working conditions based on near-infrared absorption characteristics. The method is particularly suitable for cement-based materials doped with mineral substitutes, and significantly reduces the evaluation error.

[0027] 2) By converting hyperspectral images into two-dimensional arrays, an automated and rapid retrieval method for hyperspectral image extreme values ​​and their coordinates is proposed to facilitate rapid mapping and imaging of reflectivity gradients.

[0028] The maximum and minimum reflectance values ​​of a hyperspectral image reflect the dynamic range of the pixel, and there may be slight differences in the extreme reflectance values ​​of the image at different time points. If a unified normalization process is not performed, these differences may lead to errors and inconsistencies in the quantification of moisture distribution. The present invention proposes to convert the hyperspectral image into a two-dimensional array format through the Python "Pillow" library, and use the np.max(), np.min() and np.argwhere() functions in the numpy library to quickly retrieve the maximum and minimum reflectance values ​​and their corresponding two-dimensional coordinates, thereby avoiding manual retrieval errors and improving processing efficiency and reliability. The present invention ultimately realizes a normalized hyperspectral data analysis method based on a unified extreme value benchmark, so that the moisture distribution at different time points can be quantitatively compared under the same standard, and the mapping imaging of the reflectance gradient is facilitated by the rapid positioning of extreme points.

[0029] 3) The spectral reflectance data were calibrated based on the slice weighing method to solve the quantitative relationship between moisture content and spectral characteristics. Combined with the addition averaging method, a real-time tracking and precise quantification method for the moisture distribution and evolution during the water absorption process of cement-based materials was proposed.

[0030] Determining the unit volume moisture content during the water absorption process of cement-based materials is the key to quantifying the distribution and evolution of moisture. The present invention calibrates the hyperspectral reflectance data through the slice weighing method, slices the sample after water absorption along the height direction, calculates its unit volume moisture content, and fits the functional relationship between reflectance and moisture content in combination with the spectral characteristics of the key wavelength band. The moisture distribution visualization method proposed based on the addition averaging method reduces the random noise and systematic error in a single measurement, thereby improving the signal-to-noise ratio of the spectral image and enhancing the visualization effect. The present invention establishes a real-time tracking and precise quantification method for moisture distribution based on spectral calibration, which can visualize the dynamic distribution of moisture in the water absorption process of cement-based materials in real time and significantly improve the reliability and accuracy of the measurement.

[0031] The method for rapidly assessing the water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology of the present invention has the following beneficial effects:

[0032] 1) Existing in-situ methods for assessing the depth of water penetration in physical cement-based structures have low accuracy, especially for cement-based structures doped with mineral substitutes such as blast furnace slag. The present invention uses the specific absorption characteristics of the -OH functional group in water molecules in the near-infrared band to quickly and efficiently determine the position of the wetting front and assess the water penetration depth more quickly.

[0033] 2) Computer functions are used to assist in retrieving the maximum and minimum reflectances of hyperspectral images and their corresponding coordinates, which helps to quickly map the reflectance and ensure that the hyperspectral images at different time points are established on the same basis, thereby improving the reliability and comparability of the results.

[0034] 3) The addition averaging method reduces random noise and systematic errors in the measurement, improves the signal-to-noise ratio of the spectral image, and enhances the visualization of moisture distribution. Calibration using slice weighing provides a precise quality foundation for spectral reflectance data, establishing a quantitative relationship between spectral information and actual moisture content, improving assessment accuracy and reducing costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a workflow diagram of the water absorption test experiment and hyperspectral camera scanning of cement-based material samples in the embodiment;

[0036] Figure 2 Schematic diagram of the system architecture of an infrared hyperspectral camera scanning a cement-based material sample in an embodiment;

[0037] Figure 3 The images before and after hyperspectral image correction in step 3 of the embodiment, wherein (a) the original mapping image, (b) the dark-corrected image, and (c) the white-corrected image;

[0038] Figure 4This is a flow chart showing the visualization of water penetration in the embodiment;

[0039] Figure 5 The reflectivity gradient map and partitions of the cement-based material sample during the water absorption process obtained in step 4 of the embodiment;

[0040] Figure 6 This is a time-varying characteristic diagram of the reflectivity at a wavelength of 1410 nm along the sample height direction in step 4 of the embodiment;

[0041] Figure 7 The water penetration assessment model obtained in step 4 of the embodiment;

[0042] Figure 8 Graphs showing the maximum reflectivity (a) and minimum reflectivity (b) of the cement-based material sample in step five of the embodiment;

[0043] Figure 9 The calibration line between the water content and the reflectivity obtained in step 6 of the embodiment;

[0044] Figure 10 This is the visualization of the water penetration process and water content in step seven of Example. DETAILED DESCRIPTION

[0045] Specific embodiment 1: The method for rapid assessment of water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology in this embodiment is implemented in the following steps:

[0046] Step 1: Water absorption test:

[0047] The bottom of the cement-based material sample is immersed in water to conduct a water absorption test experiment;

[0048] Step 2: Near-infrared hyperspectral camera scanning:

[0049] An infrared hyperspectral camera was installed above the cement-based material sample. During the water absorption test, the infrared hyperspectral camera was used to capture hyperspectral images of the sample at different water absorption times.

[0050] Step 3: Hyperspectral image correction:

[0051] Performing dark correction and white correction on the sample hyperspectral image obtained in step 2 to obtain a corrected sample hyperspectral image;

[0052] Step 4: Quick assessment of reflectance spectral gradient and water penetration depth:

[0053] The reflectivity gradient map of the cement-based material sample along the height direction is obtained from the corrected sample hyperspectral image. The reflectivity gradient map uses wavelength as the horizontal axis and reflectivity as the vertical axis. The reflectivity gradient curve of the cement-based material sample at different heights is drawn. The valley wavelength is determined from the reflectivity gradient curve, and the cement-based material sample is divided into a dry area, a transition area (including the wetting front), and a completely wetting area (corresponding to the water absorption time).

[0054] The reflectivity at the valley wavelength at different heights was obtained from the reflectivity gradient map of each cement-based material sample. The sample height was used as the horizontal axis and the reflectivity at the valley wavelength as the vertical axis to obtain the sample height-reflectivity curve for the corresponding water absorption time. The inflection point (mutation turning point) from the transition zone to the dry zone is the water penetration depth. The sample height-reflectivity curve for different water absorption times during the water absorption test experiment was plotted to obtain a time-varying characteristic diagram of the reflectivity along the sample height direction.

[0055] During the water absorption test, the square root of the water absorption time was used as the horizontal axis, and the water penetration depth was used as the vertical axis. A water penetration evaluation model inside the cement-based material was fitted to complete the water penetration depth evaluation.

[0056] Step 5: Reflectance spectrum mapping and visualization:

[0057] The calibrated hyperspectral image of the sample was generated and mapped using the added average analysis, and the area of ​​the cement-based material sample was selected as the area-of-interest (ROI).

[0058] The corrected sample hyperspectral image is imported into the Python function library in the form of a single-band grayscale image and converted into a two-dimensional array format. The maximum and minimum reflectance values ​​are obtained respectively. Then, the two-dimensional coordinates of the maximum and minimum reflectance values ​​are located. The extreme values ​​(maximum reflectance, minimum reflectance) are extracted as the normalized baseline and mapped on the corrected sample image to obtain the mapped hyperspectral image.

[0059] Step 6: Quantification and calibration equation of moisture content and spectral characteristics:

[0060] During the water absorption test, the cement-based material sample was cut into two halves along the height direction, and one half of the sample was cut into thin slices along the height direction. The unit volume moisture content of the thin slices was calculated, and the reflectance value at the valley wavelength and the unit volume moisture content were used to establish data points. The relationship curve between the reflectance value at the valley wavelength and the unit volume moisture content was obtained by fitting.

[0061] Step 7: Visual output of water penetration process and moisture content:

[0062] The mapped hyperspectral image of the cement-based material sample is output. The moisture content of different areas in the mapped hyperspectral image is calibrated through the relationship curve between the sample valley wavelength reflectance value and the unit volume moisture content, completing the rapid assessment of the moisture content distribution of the cement-based material.

[0063] Specific embodiment 2: This embodiment differs from specific embodiment 1 in that the cement-based material sample in step 1 is cylindrical.

[0064] Specific embodiment three: The difference between this embodiment and specific embodiment one or two is that in step two, the frame rate of the infrared hyperspectral camera is controlled to be 20-40 fps and the exposure time is 6-8 ms.

[0065] Specific embodiment 4: The difference between this embodiment and any one of specific embodiments 1 to 3 is that in step 2, during the water absorption test experiment, 8 to 12 hyperspectral images of the sample with different water absorption times are obtained by taking pictures with an infrared hyperspectral camera.

[0066] In this embodiment, 8 to 12 sample hyperspectral images are taken at intervals from the start of the water absorption test until the sample completely absorbs water.

[0067] Specific embodiment 5: The difference between this embodiment and any one of specific embodiments 1 to 4 is that in step 4, the lowest point in the wavelength range of 1400-1450 nm of the reflectivity gradient curve is determined as the valley wavelength.

[0068] Specific embodiment 6: The difference between this embodiment and any one of specific embodiments 1 to 5 is that in the reflectivity gradient curve of the cement-based material sample at different heights in step 4, the reflectivity corresponding to the valley wavelength in the transition zone has a gradient change along the height direction.

[0069] Specific embodiment seven: The difference between this embodiment and any one of specific embodiments one to six is ​​that in step five, the corrected sample hyperspectral image is imported into the Pillow module in the Python function library in the form of a single-band grayscale image and converted into a two-dimensional array format.

[0070] Specific embodiment eight: This embodiment differs from any one of specific embodiments one to seven in that in step six, when the water absorption time is 50 to 300 minutes, the cement-based material sample is cut into two halves along the height direction, and one half of the sample is cut into thin slices along the height direction.

[0071] Specific embodiment 9: This embodiment differs from any one of specific embodiments 1 to 8 in that the thickness of the thin slice sample in step 6 is 1 to 2 mm.

[0072] Specific embodiment 10: This embodiment differs from specific embodiments 1 to 9 in that the calculation formula for the unit volume moisture content of the thin slice sample in step 6 is as follows:

[0073]

[0074] Example: The method for rapidly assessing the water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology in this embodiment is implemented by the following steps:

[0075] Step 1: Scanning sample preparation and water absorption test

[0076] Ordinary Portland cement (OPC) was used to prepare cement slurry specimens, and the designed W / C (water-cement ratio) of the cement slurry was 0.4, 0.5 and 0.6 respectively. The stirred slurry was poured into a cylindrical mold with a diameter of 5 cm and a height of 10 cm. After 24 hours of pouring, the mold was removed and placed in water for curing. After curing for one year, the specimen was evenly cut into two equal parts and placed in a drying room at 40°C until the mass change within 24 hours was less than 0.1%. In order to prevent water absorption on the side, the specimen was wrapped with aluminum foil tape, exposing only the top and bottom surfaces. Before drying, the top and bottom of the specimen were trimmed by about 1 mm, and then the bottom surface of the specimen was partially immersed in water with an immersion height of 5 mm. Figure 1 shown.

[0077] Step 2: Determine the optimal parameters for near-infrared hyperspectral camera scanning

[0078] When scanning the cement slurry sample, the infrared hyperspectral camera lens is perpendicular to the sample, and the distance is adjusted so that the captured sample image occupies about 80% of the screen. The reflected light intensity of the sample image is a combination of the photon dose and dark current at a given exposure time during the scanning process. In order to minimize image blur and maintain image brightness, the maximum reflected light intensity is maintained at 80% of the maximum brightness, which is mainly achieved by adjusting the camera's frames per second (FPS) and the camera's exposure time. Among them, the exposure time determines the scanning speed. Under the premise of ensuring image quality, the exposure time should be minimized. The two parameter settings in this embodiment are shown in Table 1:

[0079] Table 1 Frames per second (FPS) and camera exposure time determination

[0080]

[0081] Step 3: Hyperspectral image correction

[0082] Spectral images usually need to be calibrated to ensure the accuracy of the spectral data. Calibration is mainly divided into dark calibration and white calibration. Calibration is performed by collecting black and white reference data and processing them. The calibration principle is based on formula (1). Dark calibration is achieved by covering the lens, and white calibration is achieved by scanning the reflectance spectrum of a high-reflectance standard white plate at the same position as the test sample. Figure 3As shown, green specks (spots of very low reflectivity) appear in the original image. These are primarily caused by dark current noise, pixel response non-uniformity, system noise, or baseline drift. After recording and subtracting the noise and offset from the dark reference data, the very low reflectivity spots disappear. White calibration compensates for variations in illumination and reflectivity across the scene, resulting in a uniform reflectivity across the entire image.

[0083]

[0084] Step 4: Rapid Assessment of Reflectance Spectral Gradient and Water Penetration Depth

[0085] The reflectivity gradient of cement-based materials along the height direction is obtained based on the specific absorption characteristics of the -OH functional group in the water molecule in the near-infrared band of 1400-1450nm, such as Figure 5 As shown. Based on the reflectivity values, three different areas are distinguished: dry area, transition area (including wetting front) and completely wet area. The transition area (including wetting front) gradually moves upward with the increase of water absorption time, which is also the key to determining the water penetration depth in cement paste. Taking the wavelength value of 1410nm corresponding to the peak point as the specific analysis target, the time-varying characteristics of the reflectivity along the height direction are obtained, and a linear model for water penetration evaluation inside the material is established based on the reflectivity gradient. The specific process is as follows:

[0086] The reflectance gradient map of the cement-based material sample along the height direction is obtained by using the corrected sample hyperspectral image ( Figure 5 ), the reflectivity gradient graph is plotted with wavelength as the abscissa and reflectivity as the ordinate, and the reflectivity gradient curves of cement-based material samples at different heights are drawn. The valley wavelength is determined to be 1410 nm from the reflectivity gradient curve, and the cement-based material sample is divided into a dry area, a transition area (including the wetting front), and a completely wetting area (corresponding to the water absorption time);

[0087] The reflectivity at the valley wavelength at different heights is obtained from the reflectivity gradient map of each cement-based material sample. The sample height is used as the horizontal coordinate and the reflectivity at the valley wavelength is used as the vertical coordinate to obtain the sample height-reflectivity curve for the corresponding water absorption time. The inflection point (mutation turning point) from the transition zone to the dry zone is the water penetration depth. The sample height-reflectivity curve at different water absorption times during the water absorption test experiment is plotted to obtain the time-varying characteristic diagram of the reflectivity along the sample height direction, as shown in FIG. Figure 6 As shown;

[0088] During the water absorption test, the square root of the water absorption time was used as the horizontal axis and the water penetration depth was used as the vertical axis to fit the water penetration evaluation model inside the cement-based material, as shown in the following figure: Figure 7 As shown;

[0089] Step 5: Reflectance Spectrum Mapping and Visualization

[0090] The calibrated hyperspectral image of the sample was generated and mapped using the added average analysis, and the area of ​​the cement-based material sample was selected as the area-of-interest (ROI).

[0091] The corrected sample hyperspectral image is imported into the Python function library as a single-band grayscale image and converted into a two-dimensional array format through the Python "Pillow" library. The maximum and minimum reflectance values ​​are obtained by using the np.max() and np.min() functions in the numpy library, respectively. Then, the two-dimensional coordinates of the maximum and minimum reflectance values ​​are located using the np.argwhere() function. Figure 8 The reflectance spectrum data at the maximum and minimum extreme points are displayed. The extreme values ​​(maximum reflectance 6592, minimum reflectance 1437) are extracted as normalized baselines and mapped onto the calibrated sample image to obtain the mapped hyperspectral image.

[0092] The visualization principle is as follows:

[0093]

[0094] In this embodiment, each unit pixel in the sample hyperspectral image is treated as a single ROI area, and the average reflectivity is equal to the reflectivity value of the unit pixel. N is the total number of gradients for the visualized color gradient. In this embodiment, the color gradient N is set to 100. Based on the above visualization principle, pseudo-color mapping is used to display the color gradient corresponding to the reflectivity values ​​of different pixels, thereby obtaining the mapped hyperspectral image.

[0095] Step 6: Quantification of moisture content and spectral characteristics and calibration equations

[0096] The cement slurry sample was evenly cut into two identical pieces along the height direction for use in the hyperspectral camera spectral scanning and moisture content calibration, respectively. The sample water absorption process was kept consistent. After the same time (300 min), the sample used for moisture content calibration was cut into 2 mm thick slices along the height direction and the unit volume moisture content was calculated according to formula (4). The relationship between the sample reflectance value and unit volume moisture content at a wavelength of 1410 nm was fitted by a functional relationship ( Figure 10 ) and use this functional relationship to calibrate the internal moisture distribution of the sample during water absorption.

[0097]

[0098] Step 7: Visual output of water penetration process and moisture content

[0099] Based on the mapped hyperspectral image, the moisture content in different areas is calibrated through the relationship curve between the valley wavelength reflectance value of the sample and the moisture content per unit volume, and the moisture penetration distribution and moisture content image of the cement-based material sample are output, such as Figure 10 As shown, a rapid assessment of the moisture content distribution of cement-based materials is completed.

[0100] This embodiment is made by Figure 10 It can be seen that dry and completely wet areas occupy the vast majority of the sample, with the wetting front maintaining an irregular line and gradually moving forward over time. The moisture distribution in the transition zone is uneven, with significant variations along the height. In particular, the moisture content at the end of the transition zone varies significantly, which is due to the complexity and variability of the local pore structure. This method enables rapid and effective assessment of the distribution and content of water permeation in cement-based materials, providing an efficient and practical research tool for the quantitative evaluation of the water permeation behavior of cement-based materials and the durability of engineering structures.

Claims

1. A rapid assessment method for moisture penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology, characterized by The rapid assessment method for moisture penetration depth and moisture content distribution of cement-based materials is implemented according to the following steps: Step 1: Water absorption test: The bottom of the cement-based material sample is immersed in water to conduct a water absorption test experiment; Step 2: Near-infrared hyperspectral camera scanning: An infrared hyperspectral camera was installed above the cement-based material sample. During the water absorption test, the infrared hyperspectral camera was used to capture hyperspectral images of the sample at different water absorption times. Step 3: Hyperspectral image correction: Performing dark correction and white correction on the sample hyperspectral image obtained in step 2 to obtain a corrected sample hyperspectral image; Step 4: Quick assessment of reflectance spectral gradient and water penetration depth: The reflectivity gradient map of the cement-based material sample along the height direction is obtained from the calibrated sample hyperspectral image. The reflectivity gradient map uses wavelength as the horizontal axis and reflectivity as the vertical axis. The reflectivity gradient curve of the cement-based material sample at different heights is drawn. The valley wavelength is determined from the reflectivity gradient curve, and the cement-based material sample is divided into a dry area, a transition area, and a completely wet area. The reflectivity at the valley wavelength at different heights was obtained from the reflectivity gradient map of each cement-based material sample. The sample height was used as the horizontal axis and the reflectivity at the valley wavelength as the vertical axis to obtain the sample height-reflectivity curve corresponding to the water absorption time. The inflection point from the transition zone to the dry zone is the water penetration depth. The sample height-reflectivity curve at different water absorption times during the water absorption test was plotted to obtain the time-varying characteristic diagram of the reflectivity along the sample height direction. During the water absorption test, the square root of the water absorption time was used as the horizontal axis, and the water penetration depth was used as the vertical axis. A water penetration evaluation model inside the cement-based material was fitted to complete the water penetration depth evaluation. Step 5: Reflectance spectrum mapping and visualization: The calibrated hyperspectral image of the sample was generated and mapped using the added average analysis, and the area of ​​the cement-based material sample was selected as the region of interest; The corrected sample hyperspectral image is imported into the Python function library in the form of a single-band grayscale image and converted into a two-dimensional array format. The maximum and minimum reflectance values ​​are obtained respectively. Then, the two-dimensional coordinates of the maximum and minimum reflectance values ​​are located, and the extreme values ​​are extracted. The extreme values ​​are the maximum and minimum reflectance values. The extreme values ​​are used as normalized baselines to map on the corrected sample image to obtain the mapped hyperspectral image. Step 6: Quantification and calibration equation of moisture content and spectral characteristics: During the water absorption test, the cement-based material sample was cut into two halves along the height direction, and one half of the sample was cut into thin slices along the height direction. The unit volume moisture content of the thin slices was calculated, and the reflectance value at the valley wavelength and the unit volume moisture content were used to establish data points. The relationship curve between the reflectance value at the valley wavelength and the unit volume moisture content was obtained by fitting. Step 7: Visual output of water penetration process and moisture content: For the hyperspectral image of the mapped cement-based material sample, the moisture content in different areas of the mapped hyperspectral image is calibrated through the relationship curve between the sample valley wavelength reflectance value and the unit volume moisture content, thereby completing the rapid assessment of the moisture content distribution of the cement-based material.

2. The method for rapid assessment of water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology according to claim 1 is characterized in that In step 1, the cement-based material sample is cylindrical.

3. The method for rapid assessment of water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology according to claim 1 is characterized in that In step 2, the frame rate of the infrared hyperspectral camera is controlled to be 20-40 fps, and the exposure time is 6-8 ms.

4. The method for rapid assessment of water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology according to claim 1 is characterized in that In step 2, during the water absorption test experiment, 8 to 12 hyperspectral images of the sample with different water absorption times are obtained by taking pictures with an infrared hyperspectral camera.

5. The method for rapid assessment of water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology according to claim 1 is characterized in that In step 4, the lowest point in the 1400-1450 nm wavelength range of the reflectivity gradient curve is determined as the valley wavelength.

6. The method for rapid assessment of water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology according to claim 1 is characterized in that In the reflectivity gradient curve of the cement-based material sample at different heights in step 4, the reflectivity corresponding to the valley wavelength in the transition region has a gradient change along the height direction.

7. The method for rapid assessment of water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology according to claim 1 is characterized in that In step five, the corrected sample hyperspectral image is imported into the Pillow module in the Python function library as a single-band grayscale image and converted into a two-dimensional array format.

8. The method for rapid assessment of water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology according to claim 1 is characterized in that In step 6, when the water absorption time is 50 to 300 minutes, the cement-based material sample is cut into two halves along the height direction, and one half of the sample is cut into thin slice samples along the height direction.

9. The method for rapid assessment of water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology according to claim 1, characterized in that The thickness of the thin slice sample in step 6 is 1~2 mm.

10. The method for rapid assessment of water penetration depth and moisture content distribution of cement-based materials based on hyperspectral imaging technology according to claim 1, characterized in that The calculation formula for the unit volume moisture content of the thin slice sample in step 6 is as follows: The unit of moisture content per unit volume is g / cm3.

Images