A mine ecological restoration effect evaluation system and method

Abstract

The present application belongs to the technical field of image processing, and particularly relates to a mine ecological restoration effect evaluation system and method. The system comprises a multi-source remote sensing data fusion part, a data processing and analysis part and an ecological restoration effect evaluation part. The multi-source remote sensing data fusion part is used for obtaining multi-source remote sensing image data of a target mine area containing multiple source images. The data processing and analysis part is used for performing multi-scale fractal analysis on the fused remote sensing image data to obtain scale-related fractal dimensions of the target mine area. The ecological restoration effect evaluation part is used for obtaining the degree of ecological restoration of the mine according to the relationship between the multi-spectral adaptive vegetation index in the historical data and the mine ecological restoration. The present application significantly improves the accuracy and comprehensiveness of the mine ecological restoration effect evaluation.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, specifically relating to a system and method for evaluating the effectiveness of mine ecological restoration. Background Technology

[0002] Mine ecological restoration refers to the restoration of ecosystems damaged by mining through engineering, biological, chemical, and management measures, restoring them to a stable, healthy, and sustainable state. With the exploitation of mineral resources, the ecological environment of mining areas has been severely damaged, resulting in vegetation destruction, soil erosion, land degradation, ecosystem imbalance, and increased environmental pollution. Therefore, how to effectively carry out mine ecological restoration has become an important issue of concern for governments and academia worldwide.

[0003] Evaluation of the effectiveness of mine ecological restoration is a crucial aspect of mine environmental management. With increasing mining activities, mining area ecosystems have suffered severe damage, leading to increasingly prominent problems such as vegetation loss, soil erosion, and water and soil loss. Implementing scientifically sound and reasonable ecological restoration measures is essential for effectively restoring the mine's ecological environment. However, assessing the effectiveness of these measures requires a precise ecological restoration effectiveness evaluation system.

[0004] Currently, remote sensing technology is widely used in ecological environment monitoring. By acquiring remote sensing images in different bands, changes in ecological elements such as surface vegetation, soil, and water bodies can be analyzed and assessed. The advantage of remote sensing technology lies in its ability to provide large-scale, continuous spatial information, making it suitable for ecological monitoring of large areas such as mines. However, existing remote sensing monitoring technologies still have some problems in evaluating the effectiveness of mine ecological restoration. First, traditional vegetation indices such as the Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI) have limitations in certain situations. NDVI, calculated using reflectance in the near-infrared and red bands, can reflect vegetation cover and health status, but its performance is poor under high and low vegetation cover conditions and is easily affected by soil background, lighting conditions, and atmospheric conditions. EVI improves upon the limitations of NDVI to some extent, but still cannot provide sufficiently accurate assessment results in the complex ecological environment of mines. Second, existing remote sensing data processing methods have shortcomings in information extraction and comprehensive analysis. Traditional single-band or multi-band analysis methods cannot fully utilize the multispectral information of remote sensing data and easily overlook the correlation and complementarity between different bands. In addition, simple linear combination and weighting processing cannot accurately reflect the complex ecological characteristics of the land surface, resulting in insufficient accuracy and reliability of the evaluation results. Summary of the Invention

[0005] Therefore, the main objective of this invention is to provide a system and method for evaluating the effectiveness of mine ecological restoration, which significantly improves the accuracy and comprehensiveness of the evaluation. The system achieves comprehensive improvements in data acquisition, processing and analysis, and effect evaluation, providing strong technical support for scientific decision-making and optimal resource allocation.

[0006] The technical solution adopted in this invention is as follows:

[0007] A system for evaluating the effectiveness of mine ecological restoration, comprising: a multi-source remote sensing data fusion section, a data processing and analysis section, and an ecological restoration effectiveness evaluation section; the multi-source remote sensing data fusion section is used to acquire multi-source remote sensing image data containing multiple source images of the target mining area, perform spatial detail enhancement and frequency domain fusion based on wavelet transform, and obtain fused remote sensing image data; the data processing and analysis section is used to perform multi-scale fractal analysis on the fused remote sensing image data to obtain the scale-related fractal dimension of the target mining area, and then based on the reflectance of each band in the fused remote sensing image data... The process involves calculating a preliminary multispectral adaptive vegetation index (MAVI). The MAVI is then adjusted based on its fractal dimension to obtain a revised MAVI. The ecological restoration effect evaluation section establishes a standard interval for the MAVI standard interval for mine ecological restoration based on the relationship between the MAVI and mine ecological restoration in historical data. Each standard interval corresponds to a degree of mine ecological restoration. The degree of ecological restoration of the mine is determined based on the standard interval of the revised MAVI.

[0008] Furthermore, the multi-source remote sensing data fusion section includes: a data acquisition unit, used to acquire multi-source remote sensing image data including high-resolution panchromatic images and multispectral images when acquiring multi-source remote sensing image data containing multiple source images of the target mining area; and a data processing unit, used to perform nonlinear enhancement on the high-frequency subband of the high-resolution panchromatic image, perform PCA transformation on the multispectral image, and then perform non-subsampled contourlet transformation on the first principal components of the high-resolution panchromatic image and the multispectral image, followed by adaptive fusion to obtain an intermediate fusion result, and then use total variational regularization to denoise and preserve the edges of the intermediate fusion result to obtain fused remote sensing image data.

[0009] Furthermore, the formula for adaptive fusion is as follows:

[0010]

[0011] ω MS =1-ω PAN ;

[0012] Among them, Fj,l These are the coefficients after fusion; the subscript j indicates the decomposition level, and l indicates the directional sub-band; this coefficient is the weighted sum of the corresponding coefficients of the high-resolution panchromatic image and the multispectral image, representing the features of the fused image at a specific scale and direction; These are the coefficients obtained after the high-resolution panchromatic image undergoes non-subsampled contourlet transform. They represent the high-frequency detail information of the high-resolution panchromatic image in the j-th layer and l-th direction. These are the coefficients obtained after the multispectral image undergoes unsampled contourlet transform; they represent the high-frequency detail information of the multispectral image in the j-th layer and l-th direction; ω PAN These are the weights assigned to the coefficients of a high-resolution panchromatic image; ω MS These are the weights assigned to the coefficients of the multispectral image; It is the variance of a high-resolution panchromatic image within a local window, reflecting the richness of detail in that region of the image. It is the variance of the multispectral image within a local window, which reflects the richness of detail in that region of the multispectral image; It is the variance of the coefficient differences, used to control the decay rate of the exponential term.

[0013] Furthermore, when using total variational regularization to denoise and preserve edges in the intermediate fusion results, the objective function is:

[0014]

[0015]

[0016] Where u represents the fused remote sensing image data; p is the location variable, a two-dimensional variable containing X-axis and Y-axis coordinates; || represents the absolute value operation; I fused This represents the intermediate fusion result; Ω is the image domain, indicating the entire region of the intermediate fusion result. The first-order gradient of the remote sensing image data u is used to capture the edge information of the image; The second gradient of the remote sensing image data u is used to capture the curvature information of the image; Let be the penalty function for the gradient. This is a smooth approximate total variation, where ∈ is a positive number less than 1 as defined; TV β (u) is the total anisotropic variation, defined as:

[0017]

[0018] in, To fuse the first-order gradient of remote sensing image data u in the X-axis direction; This is to fuse the first-order gradient of remote sensing image data u in the Y-axis direction.

[0019] Furthermore, in the data processing and analysis section, the process of performing multi-scale fractal analysis on the fused remote sensing image data to obtain the scale-related fractal dimension of the target mining area specifically includes: Let (x, y) be the pixel coordinates in the fused remote sensing image data u, where x is the X-axis coordinate and y is the Y-axis coordinate; use non-subsampled contourlet transform to perform multi-scale decomposition on the fused remote sensing image data u:

[0020] NSCT{u}={c J (x,y)}∪{d j,l (x,y):j∈[1,J],l∈[1,L j ]};

[0021] Among them, c J (x,y) is the low-frequency subband, d j,l (x,y) represents the directional subband, J is the decomposition level, and L... j It is the direction number of the j-th layer; j is the index; calculate the local multifractal spectrum τ for each directional sub-band. j,l (q,x,y):

[0022]

[0023] Where q is the multifractal moment, B ∈ (x,y) is a sphere with radius ∈ and centered at (x,y); ′ ,y ′ ) represents pixel coordinates, x ′ X-axis coordinates; y ′ The Y-axis coordinate is used; based on the local multifractal spectrum, the generalized Hearst index H(q,x,y) is calculated, and the Legendre transformation is used to calculate the multifractal spectrum and the local singularity index; then, based on the multifractal spectrum and the local singularity index, the scale-dependent fractal dimension is calculated.

[0024] Furthermore, the generalized Hearst exponent is calculated based on the local multifractal spectrum using the following formula:

[0025]

[0026] Where H(q,x,y) is the generalized Hearst index; the multifractal spectrum and local singularity index are calculated using the Legendre transformation according to the following formula:

[0027]

[0028] f(α(q,x,y),x,y)=qα(q,x,y)-qH(q,x,y)+2;

[0029] Where α(q,x,y) is the local singularity index; f(α,x,y) is the multifractal spectrum.

[0030] Furthermore, the scale-dependent fractal dimension D(s,x,y) is calculated using the following formula, based on the multifractal spectrum and the local singularity index:

[0031]

[0032] Where s is the scale parameter and α0 is the reference local singularity index.

[0033] Furthermore, using the following formula, a preliminary multispectral adaptive vegetation index (MSAVI) is calculated based on the reflectance of each band in the fused remote sensing image data. initial :

[0034]

[0035] Where, α NIR Reflectivity in the near-infrared band; α RED Reflectance in the red light band; α SWIR1 α represents the reflectivity of the shortwave infrared band 1. SWIR2 The reflectance is measured in the shortwave infrared band 2; NDVI is the standardized vegetation index, defined as: β and γ are the first and second adjustment factors, respectively; w i is the weight of the i-th band; n is the total number of bands in the fused remote sensing image data; Where λ i λ is the center wavelength of the i-th band. NIR It is the center wavelength of the near-infrared band; σ is the third adjustment factor.

[0036] Furthermore, using the following formula, the initial multispectral adaptive vegetation index is adjusted according to the fractal dimension to obtain the adjusted multispectral adaptive vegetation index MSAVI. final :

[0037]

[0038] Among them, D threshold The threshold value is the fractal dimension.

[0039] By adopting the above technical solutions, this invention achieves the following beneficial effects: Firstly, by fusing multi-source remote sensing data, this invention significantly improves the quality and information richness of image data. Traditional remote sensing data processing methods typically rely on data from a single platform or single band. This approach is often limited by spatial resolution and spectral information, making it difficult to comprehensively reflect the actual situation on the Earth's surface. This invention employs a fusion technique combining high-resolution panchromatic images and multispectral images, generating high-quality fused remote sensing image data through wavelet transform and frequency domain fusion. Wavelet transform effectively enhances the spatial details of the image, while frequency domain fusion integrates the spectral features of different bands, resulting in a fused image that possesses both high-resolution detail and rich spectral information. The application of this technology gives the system a higher starting point in the data acquisition stage, providing a solid foundation for subsequent analysis. Secondly, this invention employs a multi-scale fractal analysis method in the data processing and analysis section, greatly improving the ability to capture the ecological characteristics of mining areas. Traditional methods for evaluating the effectiveness of ecological restoration often rely on single-scale analysis, making it difficult to comprehensively reflect the complex structure and changing trends of the ecosystem. This invention employs non-subsampled contourlet transform (NSCT) to decompose fused remote sensing image data at multiple scales, obtaining low-frequency subbands and multiple directional subbands. Combined with calculations of local multifractal spectra and the generalized Hearst exponent, it comprehensively reveals the detailed changes in the image at different scales. Multi-scale fractal analysis not only reflects the complex structure of vegetation and soil but also quantifies the self-similarity and singularity of these structures, providing a strong mathematical basis for the scientific evaluation of ecological restoration effects. This invention also introduces a multispectral adaptive vegetation index (MSAVI) adjustment method based on fractal dimension, making the calculation of vegetation indices more accurate and sensitive. Traditional vegetation indices such as NDVI and EVI perform poorly under high and low vegetation cover conditions and are easily affected by soil background and lighting conditions. This invention significantly improves the sensitivity of vegetation indices to different vegetation conditions by introducing corrections based on reflectance and the standard deviation index (NDVI) in the shortwave infrared band. Meanwhile, the system adjusts the initial MSAVI based on fractal dimension, ensuring that the final MSAVI not only reflects vegetation cover and health but also integrates the complexity and restoration status of surface structure. The adjusted MSAVI more accurately reflects the actual ecological restoration status of the mining area, providing decision-makers with a more reliable evaluation basis. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the system structure of a mine ecological restoration effectiveness evaluation system provided in an embodiment of the present invention. Detailed Implementation

[0041] All features disclosed in this specification, or steps in all methods or processes disclosed herein, may be combined in any way, except for mutually exclusive features and / or steps.

[0042] Any feature disclosed in this specification (including any appended claims and abstract) may be replaced by other equivalent or similar features, unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features.

[0043] Example 1: Reference Figure 1 A system for evaluating the effectiveness of mine ecological restoration, comprising: a multi-source remote sensing data fusion section, a data processing and analysis section, and an ecological restoration effectiveness evaluation section; the multi-source remote sensing data fusion section is used to acquire multi-source remote sensing image data containing multiple source images of the target mine area, perform spatial detail enhancement and frequency domain fusion based on wavelet transform, and obtain fused remote sensing image data; the data processing and analysis section is used to perform multi-scale fractal analysis on the fused remote sensing image data to obtain the scale-related fractal dimension of the target mine area, and then based on the reflectance of each band in the fused remote sensing image data... The method involves calculating a preliminary multispectral adaptive vegetation index (MAVI). The MAVI is then adjusted based on its fractal dimension to obtain a refined MAVI. The ecological restoration effect evaluation section establishes a standard interval for the MAVI standard interval for mine ecological restoration based on the relationship between the MAVI and mine ecological restoration in historical data. Each standard interval corresponds to a degree of mine ecological restoration. The degree of ecological restoration of the mine is determined based on the MAVI standard interval in which the refined MAVI falls.

[0044] Specifically, in this system, the multi-source remote sensing data fusion component first needs to acquire multi-source remote sensing image data of the target mining area. Specifically, this data can come from different platforms such as satellite imagery, UAV imagery, and aerial imagery, covering multiple bands including visible light, near-infrared, and thermal infrared. These data each have their own characteristics in terms of spatial resolution, temporal resolution, and spectral resolution, providing information in different dimensions. For example, satellite imagery typically has a wide coverage area and a long time series, suitable for monitoring macroscopic changes; while UAV imagery has higher spatial resolution, providing more detailed information about the surface. After acquiring the multi-source data, the system needs to preprocess this data to eliminate geometric and radiometric differences between different data sources. This step includes geometric correction, radiometric correction, and atmospheric correction, aiming to align images from different data sources to the same spatial coordinate system and make their radiometric values ​​comparable. In geometric correction, geographic reference points or high-precision digital elevation models (DEMs) are typically used to ensure spatial consistency of the images; while radiometric correction uses standardization to convert the image's radiometric values ​​into ground reflectance, thereby eliminating radiometric differences between different sensors. After preprocessing, the system employs wavelet transform technology to enhance the spatial details of the multi-source remote sensing images. Wavelet transform is a multi-resolution analysis method that can effectively decompose different frequency components in an image and enhance high-frequency details. Specifically, wavelet transform can decompose an image into low-frequency and high-frequency components. The low-frequency component retains the overall structural information of the image, while the high-frequency component contains detailed information such as edges and textures. By appropriately enhancing the high-frequency component, the spatial resolution of the image can be improved, making its details clearer, which is more conducive to subsequent evaluation of ecological restoration effects.

[0045] Building upon spatial detail enhancement, the system further fuses multi-source remote sensing images in the frequency domain. Frequency domain fusion combines the frequency components of each source image, integrating the advantages of different data sources to generate high-quality fused remote sensing image data. This fusion method effectively preserves and extracts the feature information of each source image, thereby improving the overall quality of the fused image. Specifically, frequency domain fusion typically involves operations such as Fourier transform or discrete cosine transform (DCT). By performing spectral analysis on the image, the spatial information of the image is transformed into frequency domain information, and then fusion processing is performed in the frequency domain. The fused image retains the advantages of multi-source data while improving image detail and contrast, thus providing a high-quality data foundation for subsequent data processing and analysis. Through these steps, the multi-source remote sensing data fusion component can generate high-quality fused remote sensing image data, providing reliable data support for evaluating the effectiveness of mine ecological restoration. This fusion technology not only improves the spatial resolution and detail information of remote sensing data but also enhances the overall quality and information utilization rate of the data by integrating the advantages of different data sources. Compared with traditional single-source data methods, multi-source remote sensing data fusion has significant advantages, providing more comprehensive and detailed information on target mining areas, thus providing a more scientific and accurate basis for evaluating the effectiveness of ecological restoration.

[0046] Data processing and analysis is a core component of the mine ecological restoration effectiveness evaluation system. Its main task is to conduct in-depth analysis and processing of the fused multi-source remote sensing image data, thereby providing a scientific basis for assessing the effectiveness of ecological restoration. First, the system performs multi-scale fractal analysis on the fused remote sensing image data. This process aims to reveal the surface characteristics and complexity of the mining area. Fractal analysis is a mathematical method that describes the geometric complexity and spatial structure of the earth's surface by calculating the fractal dimension. In the evaluation of mine ecological restoration effectiveness, the fractal dimension can reflect the recovery status of surface vegetation and soil, as natural ecosystems often exhibit certain fractal characteristics. Through multi-scale fractal analysis, the system can obtain the fractal dimension of the target mining area at different scales, thus comprehensively assessing its degree of ecological restoration. Another important role of multi-scale fractal analysis is to help identify the details and trends of mine ecological restoration. By calculating the fractal dimension at different scales, the system can capture changes in surface details, such as changes in vegetation cover density and improvements in soil structure. This information is crucial for assessing the effectiveness of ecological restoration, as subtle changes often reveal progress that macroscopic changes cannot. Through multi-scale fractal analysis, the system can provide foundational data for subsequent vegetation index calculations, thereby improving the accuracy of the evaluation results.

[0047] After obtaining the multi-scale fractal dimension, the system integrates the reflectance of various bands in the remote sensing image data to calculate a preliminary multispectral adaptive vegetation index. The multispectral adaptive vegetation index is a vegetation index based on multispectral remote sensing data that reflects the coverage and health of surface vegetation. By calculating the ratio of reflectance in different bands, the system can preliminarily estimate the vegetation cover of the mining area. However, relying solely on the preliminary vegetation index calculation may not fully reflect the true state of ecological restoration, thus requiring further adjustment and optimization. To improve the accuracy of the multispectral adaptive vegetation index, the system introduces a fractal dimension adjustment mechanism. Specifically, the system adjusts the preliminary multispectral adaptive vegetation index based on the fractal dimension obtained from the aforementioned multi-scale fractal analysis. The fractal dimension reflects the geometric complexity of the land surface and the detailed changes in ecological restoration; therefore, by incorporating it into the vegetation index calculation, it can more accurately reflect the actual situation of mine ecological restoration. This adjustment mechanism not only improves the accuracy of the vegetation index but also, to some extent, eliminates noise and errors in the remote sensing data, making the evaluation results more reliable. After adjusting the vegetation index, the system uses the adjusted multispectral adaptive vegetation index to evaluate the ecological restoration effect. Specifically, based on the relationship between the multispectral adaptive vegetation index and the ecological restoration effect of mines in historical data, the system establishes a set of standard intervals for the multispectral adaptive vegetation index. Each standard interval corresponds to a specific degree of ecological restoration effect, such as no restoration, partial restoration, or complete restoration. By comparing the adjusted multispectral adaptive vegetation index with these standard intervals, the system can quantitatively assess the degree of ecological restoration of the mine. This standard interval-based evaluation method is highly scientific and practical, providing not only clear evaluation results but also guidance and reference for ecological restoration work.

[0048] The ecological restoration effectiveness evaluation section is the final and crucial part of the mine ecological restoration effectiveness evaluation system. Its core task is to scientifically, systematically, and comprehensively evaluate the effectiveness of mine ecological restoration using processed and analyzed data. First, this section establishes a set of standard intervals for the multispectral adaptive vegetation index (MAV) based on the relationship between MAV and mine ecological restoration effectiveness in historical data. These standard intervals, formed through the analysis and mining of a large amount of historical data, correspond to different levels of ecological restoration. Each standard interval represents a specific level of ecological restoration, from unrestored to fully restored, covering all stages of mine ecological restoration. In practice, after obtaining the adjusted MAV, the system compares it with the pre-established standard intervals. This process is similar to matching a given data point to a standard interval, allowing for a rapid assessment of the mine's ecological restoration level. This evaluation method is not only scientific but also efficient. Through this quantitative approach, the system avoids the subjectivity and uncertainty of traditional evaluation methods, providing more objective and accurate evaluation results. Furthermore, the ecological restoration effectiveness evaluation section also considers the complexity and diversity of the ecosystem. Mine ecological restoration involves not only vegetation recovery but also improvements in soil structure, hydrological characteristics, and biodiversity. Therefore, when establishing standard intervals, the system relies not only on vegetation indices but also comprehensively considers multiple ecological indicators. Through comprehensive analysis of these indicators, the system can provide a more complete evaluation of the ecological restoration effect. For example, while considering vegetation restoration, the system also assesses soil improvement, water source restoration, and biodiversity gains. In this way, through multi-dimensional and multi-indicator comprehensive evaluation, the restoration effect of the mine ecosystem can be fully reflected.

[0049] During the evaluation process, the system also incorporates spatial analysis techniques to evaluate different locations within the mining area. Mine ecological restoration is often a phased and regional process, and the restoration effects can vary significantly across different regions. Therefore, through spatial analysis of remote sensing images, the system can divide the mining area into multiple evaluation units and evaluate the restoration effect of each unit separately. This not only allows for an understanding of the overall restoration effect but also provides insights into the restoration progress of specific areas, offering precise guidance for further restoration work. During the evaluation process, the system also utilizes time series analysis methods to dynamically monitor and assess the ecological restoration effects. Ecological restoration is a long-term process, and the manifestation of restoration effects takes time. Therefore, by analyzing remote sensing data from different periods, the system can track the progress of ecological restoration and understand the changing trends of restoration effects. This dynamic monitoring not only helps to promptly identify problems in the restoration process but also provides a scientific basis for adjusting restoration strategies. For example, if the restoration effect in a certain area stagnates for a long period, the system can issue a timely warning and suggest corresponding improvement measures. It is worth noting that the ecological restoration effect evaluation also considers the impact of external factors such as climate change. The restoration of mine ecosystems depends not only on human intervention but also on natural environmental conditions. The system analyzes climate data to assess the potential impact of climate change on restoration effectiveness, ensuring the scientific rigor and reliability of the evaluation results. For example, when evaluating vegetation restoration effectiveness, the system considers changes in climate factors such as precipitation and temperature to avoid misjudgments caused by abnormal climate conditions.

[0050] Before the ecological restoration project was implemented, long-term mining operations had severely damaged surface vegetation, disrupted soil structure, depleted water resources, and drastically reduced biodiversity in a certain mining area. After the restoration project was implemented, through a series of measures including vegetation reconstruction, soil improvement, and water source restoration, the ecological environment of the area gradually improved. To scientifically evaluate the restoration effectiveness, the mining area adopted a mine ecological restoration effectiveness evaluation system.

[0051] The system first collected multi-source remote sensing image data of the mining area before and after restoration. This data included satellite imagery, UAV imagery, and aerial imagery from different time points, covering multiple bands such as visible light, near-infrared, and thermal infrared. By performing geometric, radiometric, and atmospheric corrections on this data, the system eliminated geometric and radiometric differences between different data sources, ensuring data consistency and comparability. The system employed wavelet transform technology to enhance the spatial details of the multi-source remote sensing images and fused the images in the frequency domain, generating high-quality fused remote sensing image data. This step improved the spatial resolution and detail information of the images, providing a high-quality data foundation for subsequent data processing and analysis. The system performed multi-scale fractal analysis on the fused remote sensing image data, calculating the fractal dimension of the target mining area at different scales. These fractal dimensions reflect the complexity and spatial structure characteristics of surface vegetation, providing an important reference for assessing the effectiveness of ecological restoration. In this example, the changes in fractal dimension show the gradual restoration of vegetation cover, improvement in soil structure, and increased complexity of the ecosystem in the mining area.

[0052] The system calculates a preliminary multispectral adaptive vegetation index (MAVI) based on the reflectance of each band in fused remote sensing image data. To improve the accuracy of the index, the system adjusts the preliminary MAVI according to the fractal dimension, resulting in a more accurate adjusted MAVI. These adjusted indices can more realistically reflect the actual situation of mine ecological restoration. Based on the relationship between MAVI and the effectiveness of mine ecological restoration in historical data, the system establishes a set of standard intervals for MAVI. For example, the system divides MAVI values ​​into the following intervals:

[0053] MAVI < 0.2: Not recovered

[0054] 0.2≤MAVI<0.4: Preliminary recovery

[0055] 0.4 ≤ MAVI < 0.6: Partial recovery

[0056] 0.6 ≤ MAVI < 0.8: Significant recovery

[0057] MAVI ≥ 0.8: Full recovery

[0058] By comparing the adjusted MAVI value with these standard ranges, the system can quickly determine the degree of ecological restoration in the mining area. In this example, the adjusted MAVI value is 0.65, indicating that the ecological restoration of the mining area has reached a significant level. The system further incorporates spatial analysis techniques to evaluate different regions of the mining area, finding that the restoration effect in the central and eastern areas is better, with MAVI values ​​generally above 0.7, falling into the range of significant to complete restoration; while the restoration effect in the western and northern areas is relatively poor, with MAVI values ​​between 0.4 and 0.6, falling into the range of partial restoration. These zoning evaluation results provide a scientific basis for further optimization of the restoration project.

[0059] Example 2: The multi-source remote sensing data fusion part includes: a data acquisition unit, used to acquire multi-source remote sensing image data including high-resolution panchromatic images and multispectral images when acquiring multi-source remote sensing image data containing multiple source images of the target mining area; and a data processing unit, used to perform nonlinear enhancement on the high-frequency subband of the high-resolution panchromatic image, perform PCA transformation on the multispectral image, and then perform non-subsampled contourlet transformation on the first principal components of the high-resolution panchromatic image and the multispectral image, and then perform adaptive fusion to obtain the intermediate fusion result. Finally, the intermediate fusion result is denoised and edge-preserving using total variational regularization to obtain the fused remote sensing image data.

[0060] Specifically, firstly, the system's data acquisition unit is responsible for collecting multi-source remote sensing image data of the target mining area. This data includes high-resolution panchromatic images and multispectral images. High-resolution panchromatic images have high spatial resolution and can provide detailed information about the surface; while multispectral images cover information from multiple bands and can reflect the spectral characteristics of the surface. Combining these two types of images can provide a rich source of information for the comprehensive evaluation of the mine's ecological restoration effect. Next, the data processing unit preprocesses and processes the acquired remote sensing image data. First, nonlinear enhancement is performed on the high-frequency subband of the high-resolution panchromatic image. The high-frequency subband contains edge and detail information in the image. Nonlinear enhancement can highlight these details, resulting in better clarity and resolution after image fusion. Nonlinear enhancement methods typically employ adaptive enhancement functions to avoid noise amplification caused by over-enhancement while preserving image details. While processing the high-resolution panchromatic image, the system also performs principal component analysis (PCA) transformation on the multispectral image. PCA transformation is a dimensionality reduction method that simplifies data processing complexity by transforming the information in the multispectral image into a few principal components while preserving the main information in the image. The first principal component usually contains most of the image information, so focusing on the first principal component in subsequent processing can effectively improve processing efficiency.

[0061] Then, the system performs non-subsampled contourlet transform (NSCT) on the first principal components of the high-resolution panchromatic and multispectral images. NSCT is a multi-scale, multi-directional image transformation method that can decompose images at different scales and directions, capturing details and structural features. Through NSCT transformation, the main features of the panchromatic and multispectral images can be decomposed into multiple sub-bands, providing a data foundation for subsequent adaptive fusion. After completing the NSCT transformation, the system performs adaptive fusion. The adaptive fusion method adaptively selects a fusion strategy based on the characteristics of different sub-bands to maximize the preservation of the advantageous information of each image. Specifically, adaptive fusion can dynamically adjust the fusion weights based on local image features, such as gradients and textures, so that the fused image can simultaneously retain high-resolution details and multispectral spectral information in different regions. In this way, the generated intermediate fusion result has both high-resolution details and rich spectral information, which can comprehensively reflect the ecological characteristics of the mining area. However, the intermediate result obtained by directly performing adaptive fusion may contain some noise and unnecessary details. To further improve image quality, the system uses total variational regularization to denoise and preserve edges in the intermediate fusion result. Total variational regularization (TVRF) is a commonly used method for image denoising. By minimizing the total variation of an image, it can effectively remove noise while preserving the image's edges and important details. Specifically, TVRF solves a variational problem to obtain the denoised image. During the denoising process, the image's edge and structural features are preserved, resulting in higher quality and clarity of the final fused remote sensing image data.

[0062] Example 3: The formula for adaptive fusion is as follows:

[0063]

[0064] ω MS =1-ω PAN ;

[0065] Among them, F j,l These are the coefficients after fusion; the subscript j indicates the decomposition level, and l indicates the directional sub-band; this coefficient is the weighted sum of the corresponding coefficients of the high-resolution panchromatic image and the multispectral image, representing the features of the fused image at a specific scale and direction; These are the coefficients obtained after the high-resolution panchromatic image undergoes non-subsampled contourlet transform. They represent the high-frequency detail information of the high-resolution panchromatic image in the j-th layer and l-th direction. These are the coefficients obtained after the multispectral image undergoes unsampled contourlet transform; they represent the high-frequency detail information of the multispectral image in the j-th layer and l-th direction; ω PAN These are the weights assigned to the coefficients of a high-resolution panchromatic image; ω MSThese are the weights assigned to the coefficients of the multispectral image; It is the variance of a high-resolution panchromatic image within a local window, reflecting the richness of detail in that region of the image. It is the variance of the multispectral image within a local window, which reflects the richness of detail in that region of the multispectral image; It is the variance of the coefficient differences, used to control the decay rate of the exponential term.

[0066] Specifically, the fusion coefficient F j,l Through high-resolution panchromatic image coefficients and multispectral image coefficients This is achieved through a weighted sum. The calculation of this weighted sum depends on two weight parameters ω. PAN and ω MS , where ω PAN and ω MS These represent the weights of the high-resolution panchromatic image and the multispectral image during the fusion process, respectively. The formula defines ω as follows: MS =1-ω PAN This ensures that the sum of the two weights is 1, thereby guaranteeing the stability and consistency of the fusion process. Weight ω PAN The calculation comprises two main parts: local variance ratio and difference exponential decay. The local variance ratio is determined by comparing the variances of the high-resolution panchromatic image and the multispectral image within a local window. Specifically, the weight ω... PAN It is achieved through the local variance of a high-resolution panchromatic image. Local variance of multispectral images The weight is calculated as a ratio between the two. This ratio reflects the richness of detail in that local region of the image. If the high-resolution panchromatic image has richer detail (i.e., larger local variance), its weight will increase accordingly, and vice versa. The difference exponential decay part calculates the difference between the coefficients of the high-resolution panchromatic image and the coefficients of the multispectral image, and processes this difference using an exponential decay function to further adjust the weights. Specifically, the weight ω... PAN Includes an exponential term in The coefficients representing the difference between high-resolution panchromatic images and multispectral images. This is a parameter that controls the exponential decay rate. The introduction of this exponential term reduces the weight of the high-resolution panchromatic image in regions with large coefficient differences, thereby reducing artifacts and inconsistencies that may occur in the fused image. Specifically, when the coefficients of the high-resolution panchromatic image and the multispectral image differ significantly, the value of the exponential term decreases, thus reducing ω. PAN The value of ω increases the weight of the multispectral image. Conversely, when the difference between the coefficients of the two is small, the value of the exponent term is larger. PANThe value is relatively high, and the high-resolution panchromatic image occupies a larger proportion in the fusion process. Through this adaptive adjustment mechanism, the fusion formula can dynamically balance the contributions of the two images, so that the fused image can retain rich spatial information in different regions while reflecting the true spectral characteristics.

[0067] Example 4: When using total variational regularization to denoise and preserve edges in the intermediate fusion results, the objective function is:

[0068]

[0069] Where u represents the fused remote sensing image data; p is the location variable, a two-dimensional variable containing X-axis and Y-axis coordinates; || represents the absolute value operation; I fused This represents the intermediate fusion result; Ω is the image domain, indicating the entire region of the intermediate fusion result. The first-order gradient of the remote sensing image data u is used to capture the edge information of the image; The second gradient of the remote sensing image data u is used to capture the curvature information of the image; Let be the penalty function for the gradient. This is a smooth approximate total variation, where ∈ is a positive number less than 1 as defined; TV β (u) is the total anisotropic variation, defined as:

[0070]

[0071] in, To fuse the first-order gradient of remote sensing image data u in the X-axis direction; This is to fuse the first-order gradient of remote sensing image data u in the Y-axis direction.

[0072] Specifically, in the evaluation system for the effectiveness of mine ecological restoration, Total Variation Regularization (TV) is a key image processing technique used for denoising and edge preservation of intermediate fusion results. TV transforms the denoising and edge preservation requirements into an optimization problem by constructing a complex objective function, thereby generating high-quality fused remote sensing image data. This objective function comprehensively considers multiple factors such as data fidelity, gradient penalty, second-order gradient, gradient square ratio, and total anisotropic variation. First, the data fidelity term in the objective function is expressed in integral form, i.e. Where u represents the final fused remote sensing image data, I fused It's a fusion of intermediate results. Weights This is the average weight of the high-resolution panchromatic image and the multispectral image. This part ensures the similarity between the fused image and the original fused result, allowing the denoised image to retain the features and details of the original data. By minimizing this, the optimization process can retain the feature information of the original image to the greatest extent. Secondly, the gradient penalty term... It is the core part of total variational regularization. Used for smoothing image gradients. This is the gradient of the image. The gradient penalty term penalizes the image gradient, avoiding the loss of detail caused by over-smoothing, while effectively removing noise from the image. This term is particularly suitable for preserving edge information in the image, because the gradient can capture edges and details in the image. ∈ is a positive number less than 1, used to avoid the gradient being zero, thereby improving the stability of numerical computation. Third, the second-order gradient term in the objective function. By penalizing the second-order gradient of the image, the curvature information of the image is captured, further smoothing the image while preserving its main structure. This term is particularly effective for removing high-frequency noise in the image, as high-frequency noise often manifests as large second-order gradient values. By penalizing the second-order gradient, the optimization process can generate more natural and smoother images while reducing noise. Furthermore, the gradient squared ratio term... By controlling the ratio of the squared gradients, image details are further smoothed. This term is designed to balance the trade-off between denoising and detail preservation. By minimizing this term, the system can remove noise while preserving the image's edge information to the greatest extent possible, ensuring that edge areas are not blurred due to denoising. This is particularly important in evaluating the effectiveness of mine ecological restoration, as edge information is crucial for identifying and analyzing ecological changes in mining areas. Finally, the objective function includes the anisotropic total variation term TV. β (u) measures the gradient changes of the image in the X and Y axis directions, by analyzing... and Penalties are applied to ensure smooth image processing in different directions. The total anisotropic variation is expressed as an integral. This measure, while preserving the overall image structure and edge details, reduces the impact of anisotropic noise, further improving image quality. Total variation regularization optimizes the fusion intermediate results by constructing an objective function containing multiple constraints and penalties, achieving image denoising and edge preservation. The data fidelity term ensures that the original features of the image are not destroyed, while the gradient penalty term and the second-order gradient term effectively remove noise by penalizing the image's gradient and second-order gradient, while preserving the image's edge information. The gradient squared ratio term further balances the requirements of denoising and detail preservation, while the anisotropic total variation term ensures the overall smoothness and structural integrity of the image by controlling the gradient changes in different directions.

[0073] Example 5: Data processing and analysis. The process of performing multi-scale fractal analysis on the fused remote sensing image data to obtain the scale-dependent fractal dimension of the target mining area specifically includes: Let (x, y) be the pixel coordinates in the fused remote sensing image data u, where x is the X-axis coordinate and y is the Y-axis coordinate; use non-subsampled contourlet transform to perform multi-scale decomposition on the fused remote sensing image data u:

[0074] NSCT{u}={c J (x,y)}∪{d j,l (x,y):j∈[1,J],l∈[1,L j ]};

[0075] Among them, c J (x,y) is the low-frequency subband, d j,l (x,y) represents the directional subband, J is the decomposition level, and L... j It is the direction number of the j-th layer; j is the index; calculate the local multifractal spectrum τ for each directional sub-band. j,l (q,x,y):

[0076]

[0077] Where q is the multifractal moment, B ∈ (x,y) is a sphere with radius ∈ and centered at (x,y); ′ ,y ′ ) represents pixel coordinates, x ′ X-axis coordinates; y ′ The Y-axis coordinate is used; based on the local multifractal spectrum, the generalized Hearst index H(q,x,y) is calculated, and the Legendre transformation is used to calculate the multifractal spectrum and the local singularity index; then, based on the multifractal spectrum and the local singularity index, the scale-dependent fractal dimension is calculated.

[0078] Specifically, in the mine ecological restoration effectiveness evaluation system, the multi-scale fractal analysis process in the data processing and analysis section, through refined decomposition and complex mathematical processing of fused remote sensing image data, can comprehensively reveal the geometric structure and ecological changes of the target mining area. Its core steps involve non-subsampled contourlet transform (NSCT), multi-fractal spectrum calculation, and analysis of the generalized Hearst exponent, ultimately obtaining the scale-dependent fractal dimension. These steps are supported by profound mathematical principles and image processing techniques, which will be elaborated below. First, non-subsampled contourlet transform (NSCT) is performed on the fused remote sensing image data u to achieve multi-scale decomposition of the image. NSCT is an advanced image transformation method that can effectively capture image details and edge information while maintaining spatial consistency. Through NSCT, the image is decomposed into a low-frequency subband c. J(x,y) and multiple directional subbands d j,l (x,y). The low-frequency subband mainly contains the overall structure of the image, while the directional subband reflects the high-frequency detail information of the image at different directions and scales. This decomposition method can represent the complex information of the image hierarchically, facilitating subsequent multi-scale analysis. After completing the multi-scale decomposition of the image, the system analyzes each directional subband d... j,l Local multifractal spectrum τ of (x,y) j,l The calculation of (q,x,y). The calculation of the local multifractal spectrum reveals the detail variations of an image at different scales by analyzing the distribution of pixel intensity. Specifically, the multifractal spectrum reflects the complexity and self-similarity of an image by measuring moments at different q values. Here, q is a multifractal moment representing the sensitivity to different scales. By progressively reducing the neighborhood radius ∈, the pixel intensity |d| within the neighborhood centered at (x,y) is calculated. j,l (x ′ ,y ′ )| q The cumulative sum, followed by taking the logarithmic relation, finally yields the local multifractal spectrum τ. j,l (q,x,y).

[0079] The calculation of multifractal spectra reflects the local complexity and detail variations of images at different scales. For example, in ecologically restored mining areas, different vegetation cover types and soil structures exhibit different fractal characteristics. Multifractal spectra allow for quantitative analysis of these changes, revealing the specific effects of ecological restoration. After obtaining the local multifractal spectrum, the system further calculates the generalized Hearst exponent H(q,x,y). The generalized Hearst exponent is an index used to describe data self-similarity and roughness. By calculating the generalized Hearst exponent, the changing trends of the image at different scales can be quantified. Specifically, the generalized Hearst exponent is calculated by analyzing the multifractal spectrum τ... j,l The Legendre transformation of (q,x,y) yields:

[0080]

[0081] Legendre transform can convert multifractal spectra into the generalized Hearst index and local singularity index, thereby revealing the self-similarity and detail features of images at different scales. Calculating the generalized Hearst index helps understand ecological changes in mining areas. For example, an increase in the generalized Hearst index usually indicates the restoration of vegetation cover and soil structure, as the geometry of these areas becomes more complex and richer. Conversely, if the generalized Hearst index does not change significantly, it may mean that the ecological restoration effect is not obvious. Changes in the generalized Hearst index H(q,x,y) can quantify the restoration status of the ecosystem. For example, in the process of vegetation restoration, different vegetation types and cover densities will exhibit different Hearst index values. By comparing the Hearst indexes of different regions, the effectiveness and progress of ecological restoration can be assessed, providing a scientific basis for further optimization of restoration measures. Finally, based on the generalized Hearst index and multifractal spectrum, the scale-dependent fractal dimension is systematically calculated. Fractal dimension is an important indicator for describing complex geometric morphology and can quantify the complexity and detail of images at different scales. In the evaluation of the effectiveness of mine ecological restoration, changes in fractal dimension can intuitively reflect the restoration status of the ecosystem. For example, an increase in fractal dimension signifies richer vegetation cover and more complex soil structure, indicating significant effectiveness of ecological restoration measures. Specifically, fractal dimension quantifies the recovery of vegetation and soil by analyzing detailed changes in images at different scales, providing an important quantitative indicator for the scientific evaluation of ecological restoration effects. Through this multi-scale fractal analysis method, the system can comprehensively reveal the ecological changes in mining areas, providing a scientific basis for the quantitative evaluation of ecological restoration effects. Compared to traditional single-scale analysis methods, multi-scale fractal analysis not only considers detailed changes at different scales but also provides more comprehensive and accurate evaluation results through various indicators such as the generalized Hearst exponent and fractal dimension. This is of great significance for guiding mine ecological restoration work and optimizing restoration measures. For example, if the fractal dimension of a certain area increases significantly, the system can identify that the ecological restoration effect in that area is good, and then promote the restoration measures used in that area to other areas; conversely, if the fractal dimension changes in some areas are not significant, the system can suggest that the restoration efforts need to be strengthened or the restoration strategy adjusted. The data processing and analysis section systematically performs multi-scale fractal analysis on fused remote sensing image data through non-subsampled contourlet transform, multifractal spectrum calculation, and generalized Hearst exponent calculation, obtaining the scale-dependent fractal dimension of the target mining area. This process not only reveals the complex geometric structure and ecological changes of the mining area but also provides important quantitative indicators for the scientific evaluation of ecological restoration effects. Using this method, the system can more accurately assess the effectiveness of mine ecological restoration, providing a scientific basis for optimizing and improving ecological restoration work, and promoting the sustainable development of the mining ecological environment.

[0082] Example 6: Calculate the generalized Hearst exponent based on the local multifractal spectrum using the following formula:

[0083]

[0084] Where H(q,x,y) is the generalized Hearst index; the multifractal spectrum and local singularity index are calculated using the Legendre transformation according to the following formula:

[0085]

[0086] f(α(q,x,y),x,y)=qα(q,x,y)-qH(q,x,y)+2;

[0087] Where α(q,x,y) is the local singularity index; f(α,x,y) is the multifractal spectrum.

[0088] Specifically, the core of the formula for calculating the generalized Hearst exponent H(q,x,y) lies in utilizing the local multifractal spectrum τ j,l The formula (q,x,y) is used to quantify the self-similarity and roughness of an image at different scales. In the formula, H(q,x,y) is expressed through the limit operation lim j→∞ The logarithmic operation log2 is used to capture the behavior of images at large scales, thereby revealing overall features. Specifically, molecules... and denominator These represent the accumulation of multifractal spectra at the current scale level and the previous scale level, respectively. By comparing these accumulated values, the changing trends of the image at different scales can be identified. The principle of this method is to use multifractal analysis to process image data. Multifractal analysis is a mathematical tool that can capture the complexity of images, and is particularly suitable for handling the heterogeneity in natural landscapes and ecosystems. For the evaluation system of mine ecological restoration effectiveness, this method can reveal the ecological restoration status of the mining area at different scales. For example, changes in vegetation cover and the restoration of soil structure can be reflected by the generalized Hearst index. The higher the Hearst index, the more complex and healthier the ecosystem in the area, and the better the restoration effect. After obtaining the generalized Hearst index, the system further uses the Legendre transformation to calculate the local singularity index α(q,x,y) and the multifractal spectrum f(α,x,y). The Legendre transformation is a mathematical tool borrowed from thermodynamics, used to convert the generalized Hearst index into parameters with more physical meaning. By differentiating qH(q,x,y), the local singularity index α(q,x,y) is obtained, which reflects the detailed changes in local regions of the image. This index can quantify the complexity changes within a certain region, providing support for the refined analysis of ecological restoration effects. The calculation formula for the multifractal spectrum f(α,x,y) further utilizes the relationship between α(q,x,y) and the generalized Hearst index H(q,x,y). By calculating qα(q,x,y)-qH(q,x,y)+2, the self-similarity and singularity of the image at different scales are revealed. This process refines the complexity of the image to the local scale, enabling the evaluation system to identify subtle changes in mine ecological restoration. For example, different types of vegetation cover and soil remediation effects will exhibit different characteristics in the multifractal spectrum, which can help decision-makers more accurately assess the effectiveness of restoration measures. Overall, the calculation methods of the generalized Hearst index, the local singularity index, and the multifractal spectrum provide powerful data analysis tools for the evaluation system of mine ecological restoration effectiveness. Through these mathematical methods, the system can precisely capture the progress and effects of mine ecological restoration, providing quantitative indicators to assess the effectiveness and efficiency of restoration measures. Especially when processing remote sensing image data, these methods can effectively integrate multi-source data, revealing details of ecological changes that are not directly observable to the naked eye. Through these analytical methods, the system can not only provide an evaluation of the current ecological restoration effects but also predict future restoration progress trends. This is of great significance for scientific decision-making and optimal resource allocation. For example, if the generalized Hearst index and multifractal spectrum of a certain region show a significant positive trend, the restoration measures in that region can be considered effective and can be continued or promoted to other regions. If the indicators in a certain region do not change significantly, the restoration strategy needs to be reassessed, which may require increased investment or changes in restoration methods.

[0089] Example 7: The scale-dependent fractal dimension D(s,x,y) is calculated using the following formula, based on the multifractal spectrum and the local singularity index:

[0090]

[0091] Where s is the scale parameter and α0 is the reference local singularity index.

[0092] Specifically, the core idea of ​​this formula is to weigh the multifractal spectrum under different local singularity indices through integration, ultimately obtaining a comprehensive scale-dependent fractal dimension D(s,x,y). The integral part of the formula... and In the weight function The introduction of this weighting function is crucial. This weighting function is a Gaussian function centered at the reference local singularity index α0, and the scale parameter s controls the width of the Gaussian function. The weighting function's role is to weight different local singularity indices, giving greater weight to values ​​closer to α0 and less weight to values ​​farther from α0. Through this weighting, the formula can effectively highlight those local singularity indices close to the reference value α0, thus more accurately reflecting the fractal characteristics of the image at that scale. This approach helps capture subtle structural changes in complex ecosystems, especially in mine ecological restoration, where vegetation restoration and soil improvement may vary significantly across different areas. The scale parameter s plays a vital role in evaluating the effectiveness of ecological restoration. The choice of scale parameter s directly affects the width of the weighting function, thus controlling the contribution of different local singularity indices during integration. A smaller s value means the weighting function is more concentrated, emphasizing those local singularity indices very close to α0, which helps highlight details and local features; a larger s value makes the weighting function more dispersed, able to comprehensively consider more local singularity indices, suitable for capturing overall structural changes. In the process of mine ecological restoration, by adjusting the S-value, the recovery of the ecosystem can be flexibly assessed for different analytical needs.

[0093] The multifractal spectrum f(α,x,y) in the formula is a function that measures the complexity and detail richness of an image under different local singularity indices. By multiplying the multifractal spectrum by the weighting function and integrating, the comprehensive scale-dependent fractal dimension D(s,x,y) can be calculated. This process not only considers the features of the image at a single scale but also covers changes at different scales through integration, thus providing a more comprehensive evaluation index. The mine ecological restoration effectiveness evaluation system uses the scale-dependent fractal dimension D(s,x,y) to quantify and evaluate the ecological restoration effect of mining areas. For example, areas with good vegetation restoration usually exhibit a higher fractal dimension because these areas have more complex and richer geometric structures. Conversely, areas with poor restoration effects may exhibit a lower fractal dimension. This quantitative evaluation index not only intuitively reflects the effect of ecological restoration but also helps decision-makers identify areas that need further restoration and improvement. Furthermore, by analyzing the changes in fractal dimension at different scales, the system can provide a deeper understanding of the ecological restoration situation. For example, comparing the fractal dimension of the same area at different time points can reveal the progress of ecological restoration. If the fractal dimension increases significantly over time, it indicates that the repair measures have been effective; if the fractal dimension does not change much, it may be necessary to reassess and adjust the repair strategy.

[0094] Example 8: Using the following formula, a preliminary multispectral adaptive vegetation index (MSAVI) is calculated based on the reflectance of each band in the fused remote sensing image data. initial :

[0095]

[0096] Where, α NIR Reflectivity in the near-infrared band; α REd Reflectance in the red light band; α SWIR1 α represents the reflectivity of the shortwave infrared band 1. SWIR2 The reflectance is measured in the shortwave infrared band 2; NDVI is the standardized vegetation index, defined as: β and γ are the first and second adjustment factors, respectively; w i is the weight of the i-th band; n is the total number of bands in the fused remote sensing image data;

[0097] Where λ i λ is the center wavelength of the i-th band. NIR It is the center wavelength of the near-infrared band; σ is the third adjustment factor.

[0098] Specifically, the MSAVI calculation formula includes an expression based on reflectance in the near-infrared and red light bands. The core of this expression can be seen as an improvement on traditional vegetation indices (such as NDVI). Specifically, a portion of the formula... By introducing a square root term and a linear combination, the reflectance difference between the near-infrared and red light bands was further adjusted, thus more accurately reflecting the actual condition of the vegetation. The reflectance (α) in the near-infrared band (NIR) NIR ) and reflectivity (α) in the red band (RED) RED Near-infrared (NIR) and red (LCM) reflectance are two of the most commonly used parameters in vegetation index calculations. This is because healthy vegetation has high reflectance in the near-infrared band and low reflectance in the red band. By combining the reflectance of these two bands, vegetation cover and bare land can be effectively distinguished. The complex expression described above provides higher sensitivity under different vegetation conditions, reducing errors caused by variations in soil background and light conditions. Secondly, the second part of the formula introduces a multi-band weighting mechanism to comprehensively consider the impact of reflectance in different bands on the vegetation index. The expression for this part is... Where α i w represents the reflectivity of other bands. i These are the weights for the corresponding wavebands. This weighting mechanism calculates the weights using a Gaussian distribution function, i.e. Where λ i λ is the center wavelength of the i-th band. NIR It is the center wavelength in the near-infrared band, and σ is an adjustment factor. The introduction of the Gaussian distribution function makes the weight w i The formula is more sensitive to wavelengths with center wavelengths close to the near-infrared band. This means that bands adjacent to the near-infrared band have a greater weight in vegetation index calculations because these bands have similar reflectance characteristics to the near-infrared band and can more accurately reflect vegetation conditions. In this way, the formula can integrate information from different bands, improving its sensitivity to vegetation health. Furthermore, the formula incorporates a correction term based on NDVI and the reflectance of shortwave infrared bands (SWIR1 and SWIR2). NDVI is a commonly used vegetation index, expressed as... In the formula, NDVI is adjusted using a sine function and multiplied by a regulation factor β to improve its sensitivity to different vegetation cover conditions. The specific expression is... This adjustment allows the vegetation index to more accurately reflect vegetation changes across different seasons and growth stages. The reflectance of the shortwave infrared bands (SWIR1 and SWIR2) is corrected using an exponential decay function. The expression for this part is... γ is a moderating factor. The reflectance of the shortwave infrared band reflects vegetation moisture content and soil humidity. By comparing the reflectance differences between two shortwave infrared bands, the sensitivity of the vegetation index to ecological conditions can be further improved. The formula design allows MSAVI to not only capture vegetation health but also sensitively adjust for soil moisture and other environmental factors. This multi-dimensional consideration makes MSAVI more accurate and reliable in assessing the effectiveness of mine ecological restoration. For example, under drought conditions, vegetation moisture content decreases, and the reflectance of the shortwave infrared band changes significantly. By introducing a correction term for the shortwave infrared band, the formula can more accurately reflect this change. MSAVI provides a comprehensive vegetation index for quantifying vegetation restoration in mining areas. In practical applications, the system can use MSAVI to generate detailed vegetation restoration reports, helping decision-makers scientifically assess restoration effectiveness. For example, when the MSAVI value of a certain area increases significantly, it indicates good vegetation restoration in that area, and current restoration measures can be further promoted; while for areas where the MSAVI value does not change significantly, it may be necessary to reassess the restoration strategy, increase resource investment, or adopt new technologies.

[0099] Example 9: Using the following formula, the preliminary multispectral adaptive vegetation index (MSAVI) is adjusted based on the fractal dimension to obtain the adjusted MSAVI. final :

[0100]

[0101] Among them, D threshold The threshold value is the fractal dimension.

[0102] Specifically, the initial multispectral adaptive vegetation index (MSAVI_initial) is calculated based on the reflectance of each band in the fused remote sensing image data. This index provides preliminary information about vegetation cover and health status. However, to more accurately reflect the effectiveness of ecological restoration, further adjustments using fractal dimension are needed. The fractal dimension D(s,x,y) is an important parameter describing the complexity of topography and vegetation. In evaluating the effectiveness of ecological restoration, the fractal dimension reflects the geometric complexity of vegetation and surface features. By introducing a fractal dimension threshold D... threshold This can effectively distinguish areas with different levels of recovery. When the fractal dimension is higher than a threshold, it indicates that the ecosystem structure within the area is relatively complex and rich; conversely, it may indicate that the ecosystem recovery is insufficient. The core part of the adjustment formula is a nonlinear adjustment function. This function is a sigmoid function used to correlate the fractal dimension with an adjustment factor. The sigmoid function has the following form: Its characteristic is that it can smoothly map input values ​​to a range between 0 and 1. In this formula, the input value is... By adjusting the parameters β and γ, the smoothness and inflection point positions of the function can be controlled. Specifically, when the fractal dimension D(s,x,y) is higher than the threshold D... threshold When the input value is positive, the Sigmoid function output is close to 1. This means that in areas with high ecosystem complexity, the initial MSAVI value will be amplified, reflecting higher vegetation health. Conversely, when the fractal dimension is below the threshold, the input value is negative, the Sigmoid function output is close to 0, and the initial MSAVI value will be compressed, reflecting lower vegetation health. The subtraction of 0.5 in the formula adjusts the output range of the Sigmoid function from 0 to 1 to -0.5 to 0.5. This step ensures that the center value of the adjustment factor is 0, avoiding excessive amplification or reduction of the initial MSAVI. This approach also makes the formula mathematically more symmetrical, facilitating understanding and calculation. Finally, the adjusted multispectral adaptive vegetation index MSAVI... final It is obtained by multiplying the initial MSAVI by an adjustment factor. The formula is as follows:

[0103]

[0104] This adjustment formula ensures that MSAVI more accurately reflects the ecological restoration effect of mining areas. For example, in areas with good vegetation cover, the fractal dimension is usually higher, and the adjusted MSAVI value will also be higher, thus accurately reflecting the health status of the area. Conversely, in areas with insufficient vegetation restoration, the fractal dimension is lower, and the adjusted MSAVI value will be compressed, reflecting the actual restoration situation. In this way, the adjusted MSAVI not only retains the advantages of the initial MSAVI but also incorporates the spatial information of the fractal dimension, improving the accuracy and reliability of the index. For mining ecological restoration effectiveness evaluation systems, this adjustment method can provide a more scientific and comprehensive assessment of vegetation restoration, helping decision-makers to formulate and optimize restoration strategies. For example, when the adjusted MSAVI value of a certain area increases significantly, the effectiveness of the current restoration measures can be confirmed, and these measures can be further promoted to other areas; while for areas where the adjusted MSAVI value does not change much, it is necessary to reassess and adjust the restoration strategy, which may require increased resource input or the adoption of new technologies.

[0105] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result according to substantially the same method falls within the scope of the present invention. Therefore, the scope of the present invention is defined only by the appended claims.

Claims

1. A mine ecological restoration effectiveness evaluation system, characterized in that, The system comprises: a multi-source remote sensing data fusion section, a data processing and analysis section, and an ecological restoration effect evaluation section. The multi-source remote sensing data fusion section acquires multi-source remote sensing image data of the target mining area, including multiple source images, and performs spatial detail enhancement and frequency domain fusion based on wavelet transform to obtain fused remote sensing image data. The data processing and analysis section performs multi-scale fractal analysis on the fused remote sensing image data to obtain the scale-dependent fractal dimension of the target mining area. Based on the reflectance of each band in the fused remote sensing image data, a preliminary multispectral adaptive vegetation index is calculated. The preliminary multispectral adaptive vegetation index is adjusted according to the fractal dimension. The adjusted multispectral adaptive vegetation index is obtained by calculating the number of samples. The ecological restoration effect evaluation section is used to establish a standard interval for the multispectral adaptive vegetation index for mine ecological restoration based on the relationship between the multispectral adaptive vegetation index and mine ecological restoration in historical data. Each standard interval corresponds to a degree of mine ecological restoration. The degree of ecological restoration of the mine is obtained based on the standard interval of the adjusted multispectral adaptive vegetation index. The data processing and analysis section performs multi-scale fractal analysis on the fused remote sensing image data to obtain the scale-related fractal dimension of the target mine area. Specifically, this process includes: setting... It is the fusion of remote sensing image data pixel coordinates in X-axis coordinates Y-axis coordinates; non-subsampled contourlet transform is used to fuse remote sensing image data. Perform multi-scale decomposition: ； wherein, is a low frequency subband, is a directional subband, is a decomposition level, is a directional number of the level; is a sequence index; the local multifractal spectrum is computed for each directional subband : ； wherein, is a multifractal moment, is a sphere with radius centered at ; is a pixel coordinate, is an X-axis coordinate; is a Y-axis coordinate; calculating generalized Hurst exponent based on local multifractal spectrum , transforming multifractal spectrum and local singularity exponent using Legendre transformation; calculating scale-dependent fractal dimension from multifractal spectrum and local singularity exponent; calculating scale-dependent fractal dimension from multifractal spectrum and local singularity exponent by following equation : ； wherein is a scale parameter, is a reference local strangeness exponent.

2. The mine ecological restoration effectiveness evaluation system according to claim 1, characterized in that, The multi-source remote sensing data fusion part comprises: a data acquisition unit, configured to acquire multi-source remote sensing image data including high-resolution panchromatic images and multi-spectral images when acquiring multi-source remote sensing image data of a target mine area containing multiple source images; and a data processing unit, configured to perform nonlinear enhancement on a high-frequency sub-band of the high-resolution panchromatic images, perform PCA transformation on the multi-spectral images, then perform non-subsampled contourlet transformation on first principal components of the high-resolution panchromatic images and the multi-spectral images, perform adaptive fusion, obtain a fusion intermediate result, then perform denoising and edge preservation on the fusion intermediate result by using total variation regularization, and obtain fused remote sensing image data.

3. The mine ecological restoration effectiveness evaluation system of claim 2, wherein, The formula for adaptive fusion is as follows: ； ； ； wherein, is the fused coefficient; the subscript denotes the decomposition level, denotes the directional subband; this coefficient is the weighted sum of the corresponding coefficients of the high-resolution panchromatic image and the multispectral images, and represents the characteristics of the fused image in a particular scale and direction; is the coefficient of the high-resolution panchromatic image after non-subsampled contourlet transform, which represents the high-frequency detail information of the high-resolution panchromatic image in the layer and the direction; is the coefficient of the multispectral image after non-subsampled contourlet transform, which represents the high-frequency detail information of the multispectral image in the layer and the direction; is the weight assigned to the coefficient of the high-resolution panchromatic image; is the weight assigned to the coefficient of the multispectral image; is the variance of the high-resolution panchromatic image in the local window, which reflects the richness of the details of the high-resolution panchromatic image in the region; is the variance of the multispectral image in the local window, which reflects the richness of the details of the multispectral image in the region; is the variance of the coefficient difference, which is used to control the decay rate of the exponential term.

4. The mine ecological restoration effectiveness evaluation system according to claim 3, characterized in that, When the total variation regularization is used to perform denoising and edge preservation on the fusion intermediate result, the objective function is as follows: ； in, To fuse remote sensing image data; The position variable is a two-dimensional variable containing X-axis and Y-axis coordinates; This is for absolute value operations; To integrate intermediate results; The image domain represents the entire region of the intermediate fusion result; To fuse remote sensing image data The first-order gradient is used to capture edge information of the image; Image fusion of remote sensing image data The second-order gradient is used to capture the curvature information of the image; Let be the penalty function for the gradient. This is a smooth, approximately total variation, where For positive numbers less than 1 as defined; The total variation of anisotropy is defined as: ； wherein to fuse remote sensing image data a first order gradient in the X-axis direction; to fuse remote sensing image data a first order gradient in the Y-axis direction.

5. The mine ecological restoration effectiveness evaluation system according to claim 4, characterized in that, The generalized Hurst exponent is calculated based on the local multifractal spectrum by the following formula: ； wherein, is the generalized Hurst exponent; the multifractal spectrum and the local singularity exponent are calculated using the Legendre transform by the following equation: ； ； wherein, is the local singularity exponent; is the multifractal spectrum.

6. The mine ecological restoration effectiveness evaluation system according to claim 5, characterized in that, A preliminary multi-spectral adaptive vegetation index is calculated based on reflectance of each band in the fused remote sensing image data using the following equation : ； wherein, is the reflectance of the near-infrared band; is the reflectance of the red light band; is the reflectance of the short-wave infrared 1 band; is the reflectance of the short-wave infrared 2 band; is the normalized difference vegetation index, defined as: ; and are a first adjustment coefficient and a second adjustment coefficient, respectively; is the weight of the th band; is the total number of bands in the fused remote sensing image data; wherein is the center wavelength of the th band, is the center wavelength of the near-infrared band; is a third adjustment coefficient.

7. The mine ecological restoration effectiveness evaluation system according to claim 6, characterized in that, The preliminary multi-spectral adaptive vegetation index is adjusted according to the fractal dimension to obtain an adjusted multi-spectral adaptive vegetation index using the following formula : ； wherein Df is the fractal dimension threshold value.

Images