A method for detecting loess slope disease under freeze-thaw action

Abstract

This invention provides a method for detecting diseases on loess slopes under freeze-thaw conditions, belonging to the field of loess slope disease identification technology. This invention collects sample images of loess slope soil under freeze-thaw conditions and performs spectral analysis on their texture features to identify the dominant periodic frequencies under freeze-thaw conditions. Then, it extracts texture and moisture-sensitive features from field images of the loess slope and performs periodic texture analysis based on the dominant periodic frequencies, achieving a deep characterization of freeze-thaw processes. By weightedly fusing periodic texture features with moisture change features to construct a freeze-thaw microsurface change index, it not only effectively distinguishes the different manifestations of freeze-thaw processes and conventional weathering and rain erosion in soil structure and water content, but also provides interpretable quantitative indicators of disease risk, thereby achieving graded management of disease risk and greatly improving the accuracy and sensitivity of freeze-thaw disease identification.

Description

Technical Field

[0001] This invention relates to the field of loess slope disease identification technology, specifically a method for detecting loess slope diseases under freeze-thaw conditions. Background Technology

[0002] In cold or seasonally permafrost regions such as the Loess Plateau, slope soils are constantly affected by multiple natural factors, including freeze-thaw cycles, weathering, and rain erosion, making them prone to landslides and collapses that threaten engineering safety and the regional environment. The periodic fluctuations in soil temperature and moisture content caused by freeze-thaw cycles further lead to repeated expansion, contraction, and crack development in the soil particle structure; these microstructural changes exhibit obvious periodic characteristics. In contrast, weathering and rain erosion primarily manifest as continuous erosion and material migration of the soil surface, with slow changes and a lack of clear temporal periodicity.

[0003] In the prior art, CN119066884A discloses a method and system for analyzing freeze-thaw damage of rock slopes in mines. The method includes the following steps: selecting rock slope samples from mines, conducting freeze-thaw cycle tests on the rock slope samples, and obtaining rock freeze-thaw test information; setting displacement analysis indicators based on the rock freeze-thaw test information, and obtaining the displacement changes of the rock slope under freeze-thaw cycles based on the displacement analysis indicators; analyzing the plastic damage and mine safety factor of the rock slope under freeze-thaw cycles based on the rock freeze-thaw test information; establishing a stress analysis model based on the rock freeze-thaw test information, and obtaining the stress distribution of the rock slope under freeze-thaw cycles through the stress analysis model; and analyzing the damage results of the rock slope by combining the displacement changes, plastic damage, mine safety factor, and stress distribution. Although this scheme can analyze the freeze-thaw damage of rock slopes in mines, it is difficult to distinguish the effects of freeze-thaw action and conventional weathering and rain erosion on the slope, resulting in insufficient detection accuracy and a tendency for missed or incorrect judgments.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method for detecting loess slope diseases under freeze-thaw conditions, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method for detecting loess slope diseases under freeze-thaw conditions, comprising the following steps: S1: Collect several groups of soil samples from loess slopes and conduct freeze-thaw cycle tests. Collect sample images of the soil samples periodically, extract texture features based on the sample images and perform spectral analysis to identify the dominant periodic frequency under freeze-thaw action. S2: Collect field images of loess slopes, extract texture features and moisture-sensitive features from the field images to reflect the microstructure and water content changes of the loess slope surface. S3: Based on the dominant periodic frequency under freeze-thaw action, periodic texture features are generated from the texture features of the field images to reflect the periodic impact of freeze-thaw action on loess slopes. S4: The periodic texture features and moisture-sensitive features are fused to construct a freeze-thaw microsurface change index, and the risk classification of loess slope diseases is based on this index.

[0007] Preferably, step S1 includes: S101: Calculate the gray-level co-occurrence matrix for each sample image to extract texture indicators, including but not limited to contrast, entropy, and energy. Combine the texture indicators into texture features and represent them in vector form. S102: Arrange the texture features of each sample image in chronological order to obtain the texture time-series signal, and perform discrete Fourier transform on the texture time-series signal to obtain the spectral amplitude, so as to reflect the intensity distribution of texture features at different frequencies. S103: Based on the peak detection algorithm, perform spectral analysis on the spectral amplitude, identify the significant peak frequencies in the spectrum, and statistically analyze the significant peak frequencies corresponding to all sample images; S104: Calculate normalized weights based on the image quality of each sample image, use the normalized weights to weight the corresponding significant peak frequencies, and take the significant peak frequency with the largest weight as the final dominant period frequency.

[0008] Preferably, the image quality includes, but is not limited to, sharpness indicators, noise indicators, and uniformity indicators, wherein the uniformity indicators include brightness uniformity and contrast uniformity. The logic for calculating normalized weights using image quality is as follows: The Laplacian filter is used to detect the sample images, the corresponding Laplacian operator response is calculated, and the response result is used as a sharpness index. The sample image is divided into several groups of pixel blocks to generate noise and uniformity indices. The local variance method is used to process the segmented sample images, calculate the pixel variance of each group of pixel blocks, label the pixel blocks with pixel variance below a preset threshold as uniform blocks, and use the mean of pixel variance of all uniform blocks as a noise index. Calculate the overall brightness variance of the sample image and the standard deviation of the contrast of all pixel blocks, and use them as indicators of brightness uniformity and contrast uniformity, respectively. The image quality score is obtained by summing the normalized sharpness, noise, and uniformity indices. The ratio between the image quality score of each sample image and the sum of the image quality scores of all sample images is used as the normalization weight.

[0009] Preferably, the moisture-sensitive feature is obtained by detecting the near-infrared reflectance of each pixel in the field image, specifically the absolute difference in near-infrared reflectance of each pixel at adjacent time points.

[0010] Preferably, step S3 includes: S301: Perform a difference operation on the texture feature vectors of the real image at adjacent time points to obtain a texture difference vector; S302: Arrange the texture difference vectors in time order to obtain the temporal difference signal; S303: Perform a discrete Fourier transform on the time-series differential signal to obtain the spectral amplitude, and use the amplitude corresponding to the dominant periodic frequency in the spectral amplitude as the periodic texture feature.

[0011] Preferably, the freeze-thaw microsurface variation index is obtained by weighting periodic texture features and moisture-sensitive features, and the larger the freeze-thaw microsurface variation index, the stronger the impact of freeze-thaw action on loess slopes.

[0012] Preferably, the disease risk is divided into three levels: high, medium, and low, and the risk level is determined by the freeze-thaw microscopic surface change index and a preset index threshold.

[0013] Preferably, the index threshold includes a first threshold and a second threshold, which are sorted from smallest to largest; When the freeze-thaw microsurface change index is less than or equal to the first threshold, the freeze-thaw effect on loess slopes is considered weak, and the disease risk level is low. When the freeze-thaw microsurface change index is greater than the first threshold and less than or equal to the second threshold, it is considered that the freeze-thaw effect on loess slopes is enhanced and the disease risk level is medium. When the freeze-thaw microsurface change index is greater than the second threshold, the freeze-thaw effect on loess slopes is considered strong, and the disease risk level is high.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention acquires sample images of loess slope soil under freeze-thaw conditions and performs spectral analysis on their texture features to identify the dominant periodic frequencies under freeze-thaw conditions. It then extracts texture and moisture-sensitive features from field images of the loess slopes and performs periodic texture analysis based on the dominant periodic frequencies, achieving a deep characterization of freeze-thaw processes. By weightedly fusing periodic texture features with moisture change features to construct a freeze-thaw microsurface change index, it not only effectively distinguishes the different manifestations of freeze-thaw processes and conventional weathering and rain erosion in soil structure and water content, but also provides interpretable quantitative indicators of disease risk, thereby enabling graded management of disease risk and greatly improving the accuracy and sensitivity of identifying freeze-thaw diseases. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a schematic diagram of the method flow for step S1 of the present invention; Figure 3 This is a schematic diagram of the method flow for step S3 of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0017] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0018] Example: Please see Figures 1-3 The present invention provides a technical solution: A method for detecting loess slope diseases under freeze-thaw conditions, comprising the following steps: S1: Collect several groups of soil samples from loess slopes and conduct freeze-thaw cycle tests. Collect sample images of the soil samples periodically, extract texture features based on the sample images and perform spectral analysis to identify the dominant periodic frequencies under freeze-thaw action.

[0019] Step S1 includes: S101: Calculate the gray-level co-occurrence matrix for each sample image to extract texture indicators, including but not limited to contrast, entropy, and energy. Combine the texture indicators into texture features and represent them in vector form. S102: Arrange the texture features of each sample image in chronological order to obtain a texture time-series signal. Perform a Discrete Fourier Transform (DFT) on the texture time-series signal to obtain the spectral amplitude, which reflects the intensity distribution of texture features at different frequencies. The function expression for the DFT is: In the formula Represents frequency Spectral amplitude at that location This represents the Fourier operator. Indicates the first Texture feature vector of a sample image, index Indicates the index of the sample image. This represents the total number of sample images.

[0020] S103: Based on the peak detection algorithm, perform spectral analysis on the spectral amplitude, identify the significant peak frequencies in the spectrum, and statistically analyze the significant peak frequencies corresponding to all sample images; S104: Calculate normalized weights based on the image quality of each sample image, use the normalized weights to weight the corresponding significant peak frequencies, and take the significant peak frequency with the largest weight as the final dominant period frequency.

[0021] It is understandable that freeze-thaw cycles are essentially a periodic process of soil temperature and moisture content changing over time. During freeze-thaw cycles, soil structure undergoes repeated microscopic changes such as expansion, contraction, and crack formation and closure. When conducting freeze-thaw cycle tests on slope soil samples, the "freezing" and "thawing" processes are typically performed alternately at certain intervals. For example, a sample image is acquired after each complete freeze-thaw cycle (e.g., 6 hours of freezing + 6 hours of thawing). Texture features (such as contrast, entropy, and energy) are extracted from the images acquired at each moment using methods such as gray-level co-occurrence matrix. Since these features are closely related to the microscopic structure of the soil, surface structure changes at different freeze-thaw stages can be obtained through multi-temporal acquisition. Assuming that sample images are acquired at 10 time points throughout the freeze-thaw test, and texture features are extracted at each time point, these 10 sets of features represent the microscopic structure of the sample at different stages of the freeze-thaw cycle. Sorting these 10 sets of features according to the acquisition time yields a complete temporal texture signal.

[0022] Freeze-thaw cycles are driven by changes in ambient temperature, which exhibits a clear periodicity (e.g., diurnal temperature variation, seasonal changes). Soil temperature and moisture content also show periodicity, leading to periodic microscopic changes in soil structure. In contrast, ordinary weathering and erosion are primarily caused by the long-term effects of wind and runoff, resulting in surface erosion. This process tends to be slow and gradual, with more uniform surface texture and more continuous changes without obvious periodicity. Therefore, the dominant periodic frequency can be used to reflect the main frequency components reflecting the periodic changes in soil microstructure over time during freeze-thaw cycles. Using this frequency as a benchmark, periodic texture features related to freeze-thaw processes can be accurately extracted, thereby improving the targeting and accuracy of disease detection.

[0023] Image quality includes, but is not limited to, sharpness metrics, noise metrics, and uniformity metrics. Uniformity metrics include brightness uniformity and contrast uniformity. The logic for calculating normalized weights using image quality is as follows: The Laplacian filter is used to detect the sample images, the corresponding Laplacian operator response is calculated, and the response result is used as a sharpness index. The sample image is divided into several groups of pixel blocks to generate noise and uniformity indices. The local variance method is used to process the segmented sample images, calculate the pixel variance of each group of pixel blocks, and label the pixel blocks with pixel variance below a preset threshold as uniform blocks. The mean of the pixel variance of all uniform blocks is used as a noise index. This is because image noise usually manifests as random changes in local pixel gray values. Such fluctuations are more obvious in areas that are originally flat and have less texture. A uniform block is an image block with a pixel variance below a preset threshold, indicating that the area itself should be relatively uniform and flat, and theoretically should not have obvious texture changes. In other words, the gray value fluctuations in such areas mainly come from noise. Therefore, the pixel variance of uniform blocks can more accurately reflect the noise level. Calculate the overall brightness variance of the sample image and the standard deviation of the contrast of all pixel blocks, and use them as indicators of brightness uniformity and contrast uniformity, respectively. Sharpness metrics reflect image detail and edge sharpness, noise metrics reflect the intensity of random interference in the image, and uniformity metrics measure the spatial consistency of brightness and contrast. Together, these three metrics constitute a comprehensive evaluation system for image quality. Therefore, after normalizing the sharpness, noise, and uniformity metrics, the image quality score is obtained by summing them. The ratio between the image quality score of each sample image and the sum of the image quality scores of all sample images is used as the normalization weight.

[0024] Since reference images are often lacking in the image quality assessment of soil images under freeze-thaw conditions, a referenceless evaluation method is preferable. This step, by combining a multi-dimensional comprehensive evaluation method that considers Laplacian filter sharpness, local variance noise index, and brightness and contrast uniformity index, ensures that the spectral analysis is based primarily on high-quality images, avoiding false peaks or frequency shifts introduced by noise or blurred images, and greatly enhancing the accuracy and stability of freeze-thaw dominant period frequency identification.

[0025] S2: Collect field images of loess slopes, extract texture features and moisture-sensitive features from the field images to reflect the microstructure and water content changes of the loess slope surface.

[0026] Moisture-sensitive features are obtained by detecting the near-infrared reflectance of each pixel in the field image, specifically the absolute difference in near-infrared reflectance of each pixel at adjacent time points.

[0027] Understandably, changes in soil moisture are often a leading indicator of structural changes triggered by freeze-thaw cycles. Combining this with textural changes allows for earlier and more accurate detection of disease occurrence and development trends. Therefore, incorporating moisture-sensitive features as a component of the index construction, combined with periodic textural features, can jointly reveal the comprehensive impact of freeze-thaw cycles on slopes, thereby enhancing the physical representativeness and explanatory power of the index. Furthermore, since the near-infrared band is highly sensitive to moisture content, an increase in soil moisture significantly reduces near-infrared reflectance. Therefore, this parameter can effectively reflect the micro-dynamic changes in soil moisture content. In other words, the difference in near-infrared reflectance between adjacent moments can be considered a direct reflection of moisture changes, thus capturing the periodic changes and dynamic trends in moisture content during freeze-thaw cycles.

[0028] S3: Based on the dominant periodic frequency under freeze-thaw action, periodic texture features are generated from the texture features of the field images to reflect the periodic impact of freeze-thaw action on loess slopes.

[0029] Step S3 includes: S301: Perform a difference operation on the texture feature vectors of the real image at adjacent time points to obtain a texture difference vector; S302: Arrange the texture difference vectors in time order to obtain the temporal difference signal; S303: Perform a discrete Fourier transform on the time-series differential signal to obtain the spectral amplitude, and use the amplitude corresponding to the dominant periodic frequency in the spectral amplitude as the periodic texture feature.

[0030] This step is similar to the analysis method used to identify the dominant periodic frequency, except that a differential operation is added. This is because by differentiating the texture features between adjacent time steps, the changing parts of the texture can be effectively highlighted, the static components of the texture features (such as constant background texture) can be weakened, and the representation of the small, periodic structural changes caused by the freeze-thaw process in the signal can be enhanced.

[0031] When identifying the dominant periodic frequency, the overall texture time-series signal of the freeze-thaw test sample is used to obtain the frequency component distribution to reflect the inherent periodicity of the freeze-thaw process. It is necessary to retain the overall information of the signal (including low-frequency components) to accurately reveal the complete spectral structure; therefore, differential analysis is not performed to avoid losing important low-frequency information. However, during periodic analysis, the texture signal contains a large amount of non-periodic background and trends. Direct Fourier analysis may be interfered with by non-periodic changes. Therefore, differential operations are needed to highlight the time-series variation and remove static and slowly changing components, making the periodic signal more prominent. This facilitates the accurate extraction of periodic texture features induced by freeze-thaw processes, thereby better distinguishing damage caused by freeze-thaw processes (which are periodic) from conventional weathering and erosion (which are non-periodic), and enhancing the ability to identify damage under freeze-thaw conditions.

[0032] S4: The periodic texture features and moisture-sensitive features are fused to construct a freeze-thaw microsurface change index, and the risk classification of loess slope diseases is based on this index.

[0033] The freeze-thaw microsurface variation index is obtained by weighting periodic texture features and moisture-sensitive features, and the larger the freeze-thaw microsurface variation index, the stronger the impact of freeze-thaw on loess slopes. Specifically, it involves standardizing the amplitude corresponding to the dominant periodic frequency in the spectral amplitude (periodic texture feature) and the absolute difference of the near-infrared reflectance of each pixel at adjacent time points (moisture-sensitive feature), removing the dimensions, and then weighting the result. This is because freeze-thaw cycles cause the loosening of the structure between soil particles, the generation and expansion of cracks, and these changes in microstructure cause periodic changes in surface texture. At the same time, the freeze-thaw process is also accompanied by the freezing and thawing of water, affecting the physical properties and stability of loess slopes. Therefore, the freeze-thaw microsurface variation index, by weighting and fusing periodic texture features and moisture-sensitive features, can capture both the microstructural changes in surface texture caused by freeze-thaw and the dynamic changes in the water content of loess slopes, thus comprehensively reflecting the degree of impact of the freeze-thaw process on the microenvironment of loess slopes, and distinguishing areas severely affected by freeze-thaw from non-freeze-thaw or weakly freeze-thawed areas. Regarding weighting, the sum of the weights for periodic texture features and moisture-sensitive features is 1. Initially, both can be set to 0.5 (i.e., equal weighting). Subsequently, if a particular feature is more representative in describing the impact of freeze-thaw cycles, it should be assigned a higher weight. Specifically, statistical correlation analysis (such as Pearson and Spearman coefficients) can be used to determine the correlation coefficients between the two features and target variables such as the severity of freeze-thaw damage, and weights can be allocated according to the correlation. Alternatively, principal component analysis or factor analysis can be used to incorporate periodic texture and moisture-sensitive features into multivariate analysis, extract principal components, and determine weight allocation based on contribution rates. Machine learning algorithms can also be used for automatic adjustment, with the specific adjustment method determined based on expert experience.

[0034] Disease risk is divided into three levels: high, medium, and low. The risk level is determined by the freeze-thaw microscopic surface change index and the preset index threshold.

[0035] The index thresholds include a first threshold and a second threshold, which are sorted from smallest to largest. When the freeze-thaw microsurface change index is less than or equal to the first threshold, the freeze-thaw effect on loess slopes is considered weak, and the disease risk level is low. When the freeze-thaw microsurface change index is greater than the first threshold and less than or equal to the second threshold, it is considered that the freeze-thaw effect on loess slopes is enhanced and the disease risk level is medium. When the freeze-thaw microsurface change index is greater than the second threshold, the freeze-thaw effect on loess slopes is considered strong, and the disease risk level is high.

[0036] Since the freeze-thaw microsurface variation index is directly proportional to the degree of freeze-thaw impact on loess slopes, its value can be used as a quantitative indicator to assess the level of risk. The classification results can serve as a theoretical basis for subsequent maintenance adjustments and strategy formulation, thus forming a complete management loop. For the index thresholds, statistical analysis (such as box plot analysis, cluster analysis, ROC curve analysis, etc.) can be used to find the optimal dividing point to distinguish different risk levels. For example, the first and second thresholds can be determined by maximizing the difference between categories or the balance between sensitivity and specificity. Alternatively, they can be determined by combining expert experience and engineering data.

[0037] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0038] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0039] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0040] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for detecting loess slope diseases under freeze-thaw conditions, characterized in that, The specific steps include: S1: Collect several groups of soil samples from loess slopes and conduct freeze-thaw cycle tests. Collect sample images of the soil samples periodically, extract texture features based on the sample images and perform spectral analysis to identify the dominant periodic frequency under freeze-thaw action. S2: Collect field images of loess slopes, extract texture features and moisture-sensitive features from the field images to reflect the microstructure and water content changes of the loess slope surface. S3: Based on the dominant periodic frequency under freeze-thaw action, periodic texture features are generated from the texture features of the field images to reflect the periodic impact of freeze-thaw action on loess slopes. S4: The periodic texture features and moisture-sensitive features are fused to construct a freeze-thaw microsurface change index, and the risk classification of loess slope diseases is based on this index.

2. The method for detecting loess slope diseases under freeze-thaw action according to claim 1, characterized in that: Step S1 includes: S101: Calculate the gray-level co-occurrence matrix for each sample image to extract texture indicators, including but not limited to contrast, entropy, and energy. Combine the texture indicators into texture features and represent them in vector form. S102: Arrange the texture features of each sample image in chronological order to obtain the texture time-series signal, and perform discrete Fourier transform on the texture time-series signal to obtain the spectral amplitude, so as to reflect the intensity distribution of texture features at different frequencies. S103: Based on the peak detection algorithm, perform spectral analysis on the spectral amplitude, identify the significant peak frequencies in the spectrum, and statistically analyze the significant peak frequencies corresponding to all sample images; S104: Calculate normalized weights based on the image quality of each sample image, use the normalized weights to weight the corresponding significant peak frequencies, and take the significant peak frequency with the largest weight as the final dominant period frequency.

3. The method for detecting loess slope diseases under freeze-thaw action according to claim 2, characterized in that: The image quality includes, but is not limited to, sharpness indicators, noise indicators, and uniformity indicators, wherein the uniformity indicators include brightness uniformity and contrast uniformity. The logic for calculating normalized weights using image quality is as follows: The Laplacian filter is used to detect the sample images, the corresponding Laplacian operator response is calculated, and the response result is used as a sharpness index. The sample image is divided into several groups of pixel blocks to generate noise and uniformity indices. The local variance method is used to process the segmented sample images, calculate the pixel variance of each group of pixel blocks, label the pixel blocks with pixel variance below a preset threshold as uniform blocks, and use the mean of pixel variance of all uniform blocks as a noise index. Calculate the overall brightness variance of the sample image and the standard deviation of the contrast of all pixel blocks, and use them as indicators of brightness uniformity and contrast uniformity, respectively. The image quality score is obtained by summing the normalized sharpness, noise, and uniformity indices. The ratio between the image quality score of each sample image and the sum of the image quality scores of all sample images is used as the normalization weight.

4. The method for detecting loess slope diseases under freeze-thaw action according to claim 1, characterized in that: The moisture-sensitive feature is obtained by detecting the near-infrared reflectance of each pixel in the field image, specifically the absolute difference in near-infrared reflectance of each pixel at adjacent time points.

5. The method for detecting loess slope diseases under freeze-thaw action according to claim 2, characterized in that: Step S3 includes: S301: Perform a difference operation on the texture feature vectors of the real image at adjacent time points to obtain a texture difference vector; S302: Arrange the texture difference vectors in time order to obtain the temporal difference signal; S303: Perform a discrete Fourier transform on the time-series differential signal to obtain the spectral amplitude, and use the amplitude corresponding to the dominant periodic frequency in the spectral amplitude as the periodic texture feature.

6. The method for detecting loess slope diseases under freeze-thaw action according to claim 1, characterized in that: The freeze-thaw microsurface variation index is obtained by weighting periodic texture features and moisture-sensitive features, and the larger the freeze-thaw microsurface variation index, the stronger the impact of freeze-thaw on loess slopes.

7. The method for detecting loess slope diseases under freeze-thaw action according to claim 6, characterized in that: The disease risk is divided into three levels: high, medium, and low. The risk level is determined by the freeze-thaw microscopic surface change index and the preset index threshold.

8. The method for detecting loess slope diseases under freeze-thaw action according to claim 7, characterized in that: The index threshold includes a first threshold and a second threshold, which are sorted from smallest to largest. When the freeze-thaw microsurface change index is less than or equal to the first threshold, the freeze-thaw effect on loess slopes is considered weak, and the disease risk level is low. When the freeze-thaw microsurface change index is greater than the first threshold and less than or equal to the second threshold, it is considered that the impact of freeze-thaw on loess slopes is enhanced, and the disease risk level is medium. When the freeze-thaw microsurface change index is greater than the second threshold, the freeze-thaw effect on loess slopes is considered strong, and the disease risk level is high.

Images