Underground space entrance and exit identification method and system based on multi-source data fusion

Abstract

The invention relates to the technical field of remote sensing target extraction, in particular to an underground space entrance and exit identification method and system based on multi-source data fusion, and the method comprises the steps: firstly, processing optical image data through employing a mixed pixel decomposition algorithm, removing vegetation information, enhancing the spectral features of soil, and calculating the reflectivity difference of each pixel; processing the thermal infrared data, and calculating the temperature difference of each pixel; and finally, carrying out linear fusion on the reflectivity difference and the temperature difference of each pixel to generate a pixel-level fusion result, and finally generating a spatial distribution diagram of the underground space entrance and exit. According to the method, high-precision and high-efficiency underground space entrance and exit automatic identification is realized by fusing mixed pixel decomposition enhanced soil spectrum abnormity and thermal infrared temperature abnormity.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing target extraction, and in particular to a method and system for identifying underground space entrances and exits based on multi-source data fusion. Background Art

[0002] Underground spaces are diverse and encompass a wide range of types, including warehouses, factories, command posts, communication hubs, first aid stations, water supply and drainage pipelines, and heat and power pipelines. These diverse types of underground spaces not only improve land use efficiency in urban development but also enhance cities' disaster preparedness and infrastructure stability. Underground entrances and exits are crucial for underground space planning, urban safety, and emergency preparedness. Accurately monitoring underground entrances and exits plays an irreplaceable role in ensuring urban safety, efficient operation, and sustainable development.

[0003] Currently, the most effective method for monitoring underground entrances and exits is manual sampling. However, this method is time-consuming and labor-intensive, making it difficult to apply to large-scale underground space information extraction. Using remote sensing technology to accurately identify underground entrances and exits, and subsequently analyzing their structure, is a key approach for large-scale underground space monitoring. Radar waves can penetrate clouds, dust, vegetation, and other media to detect the unique shapes and structures of underground entrances and exits. However, data acquisition is challenging, processing cycles are lengthy, and the system is subject to significant electromagnetic interference from power lines, mobile phone signals, radio antennas, and other environmental factors. Optical remote sensing analyzes multispectral or hyperspectral remote sensing imagery, exploiting the differences in spectral reflectance between underground entrances and exits and surrounding objects such as vegetation, soil, and buildings to achieve identification. Due to factors such as air circulation and human activity, the temperature of underground entrances and exits differs from that of the surrounding environment. Thermal infrared remote sensing imagery can capture these temperature differences and thus identify entrances and exits. Although existing methods have achieved the identification of underground entrances and exits, most rely on feature analysis of single data points, and their accuracy remains to be improved. Summary of the Invention

[0004] The present invention aims to solve the problem that the existing technology relies on a single data source (such as optical or thermal infrared remote sensing) and fails to effectively eliminate vegetation interference, resulting in low accuracy in identifying underground space entrances and exits and poor resistance to environmental interference. A method and system for identifying underground space entrances and exits based on multi-source data fusion is proposed to extract underground space entrances and exits in the image coverage area, and further improve the accuracy of underground space entrance and exit extraction by considering feature enhancement technology and multi-source data fusion technology.

[0005] In order to achieve the above purpose, the technical solutions adopted are:

[0006] The present invention provides a method for identifying underground space entrances and exits based on multi-source data fusion, comprising the following steps:

[0007] Step 1: Use the hybrid pixel decomposition algorithm to process the optical image data, remove vegetation information and enhance soil spectral characteristics, and calculate the reflectance difference of each pixel;

[0008] Step 2: Process the thermal infrared data and calculate the temperature difference of each pixel;

[0009] Step 3: Linearly fuse the reflectivity difference and temperature difference of each pixel to generate a pixel-level fusion result, and finally generate a spatial distribution map of the entrances and exits of the underground space.

[0010] According to the underground space entrance and exit identification method based on multi-source data fusion of the present invention, step 1 specifically includes:

[0011] Determine the vegetation spectral range based on historical empirical data and use it as a constraint to constrain the mixed pixel decomposition results;

[0012] The sliding window size is determined according to the number of mixed pixel bands of the image data, and the image data is processed in blocks;

[0013] The soil endmember reflectance of all pixels in each window is extracted using a mixed pixel decomposition algorithm, and the difference between the soil reflectance of each pixel and the window mean soil reflectance is calculated. The window mean soil reflectance refers to the average value of the soil endmember reflectance extracted after mixed pixel decomposition of all pixels in the sliding window.

[0014] According to the underground space entrance and exit identification method based on multi-source data fusion of the present invention, the soil end-member reflectance of all pixels in each window is extracted using a hybrid pixel decomposition algorithm, and the difference between the soil reflectance of each pixel and the window soil mean reflectance is calculated, specifically including:

[0015] In mixed pixel decomposition, the spectrum of each pixel is expressed as:

[0016]

[0017] Among them, R leaf and R soil are vegetation and soil reflectance, α and β are proportional coefficients, and ε is the error value;

[0018] There are N pixels in each sliding window, and a total of M×(N+4) equations are generated, where M is the number of bands, and 4 is the four constraint equations established based on the lower and upper limits of soil reflectance and vegetation reflectance in each band. α, β, and R are solved by the least squares method. leaf and R soil ;

[0019] Output the soil end member reflectance R of all pixels in the sliding windowsoil , calculate the mean soil reflectance of all pixels in the sliding window

[0020] Calculate the soil reflectance R for each pixel soil and the mean reflectivity of the window soil difference.

[0021] According to the underground space entrance and exit identification method based on multi-source data fusion of the present invention, further, the calculation formula of the reflectivity difference is:

[0022]

[0023] Among them, B i is the soil reflectance of the current pixel in the i-th band, is the mean soil reflectance of all pixels in the sliding window, and M is the maximum difference value in the sliding window.

[0024] According to the underground space entrance and exit identification method based on multi-source data fusion of the present invention, step 2 specifically includes:

[0025] Perform atmospheric correction and radiation calibration on thermal infrared data and convert them into surface temperature values;

[0026] The spatial resolution of thermal infrared data was adjusted to be consistent with that of optical image data using Kriging spatial interpolation method;

[0027] The data is divided into blocks based on a sliding window and the temperature difference is calculated block by block.

[0028] According to the underground space entrance and exit identification method based on multi-source data fusion of the present invention, further, the temperature difference calculation formula is:

[0029]

[0030] Among them, H i is the temperature value of the current pixel, is the average temperature value of all pixels in the sliding window, and H is the maximum temperature difference value in the sliding window.

[0031] According to the underground space entrance and exit identification method based on multi-source data fusion of the present invention, further, in step 3, the reflectivity difference and temperature difference of each pixel are linearly fused to generate a pixel-level fusion result, which is expressed as:

[0032] D result =xD opt +yD the

[0033] Among them, D opt is the soil reflectivity, Dthe is the temperature difference, x and y are decision fusion parameters, and x+y=1.

[0034] According to the underground space entrance and exit identification method based on multi-source data fusion of the present invention, further, the values of the decision fusion parameters x and y are dynamically adjusted according to the size of the temperature difference.

[0035] According to the underground space entrance and exit identification method based on multi-source data fusion of the present invention, further, generating a spatial distribution map of underground space entrances and exits in step 3 includes:

[0036] The extreme points of the fused data are extracted based on the sliding window to determine the specific spatial locations of the entrances and exits of the underground space;

[0037] Move the position of the sliding window and repeat the above steps until all image data are processed, and extract the spatial locations of the entrances and exits of the underground space in the image data.

[0038] Furthermore, the present invention also provides an underground space entrance and exit identification system based on multi-source data fusion, which is used to implement the above-mentioned underground space entrance and exit identification method based on multi-source data fusion, comprising:

[0039] The reflectance difference calculation module is used to process the optical image data using a hybrid pixel decomposition algorithm, remove vegetation information and enhance soil spectral characteristics, and calculate the reflectance difference of each pixel;

[0040] Temperature difference calculation module, used to process thermal infrared data and calculate the temperature difference of each pixel;

[0041] The multi-source data fusion module is used to linearly fuse the reflectivity difference and temperature difference of each pixel to generate pixel-level fusion results, and finally generate a spatial distribution map of the entrances and exits of the underground space.

[0042] The beneficial effects achieved by adopting the above technical solution are:

[0043] 1. High recognition accuracy. Existing methods for extracting underground space entrances and exits at the pixel level only consider the abnormal characteristics of the image pixel spectrum without performing any enhancement processing. The spectral signal is easily affected by the environment, which increases the difficulty of differentiation. This invention reduces the influence of the surrounding environment through feature enhancement, increases the dimension of the identification features of underground space entrances and exits, and effectively improves the recognition accuracy.

[0044] 2. High computational efficiency. The algorithm analyzes and processes optical and thermal infrared remote sensing data in parallel, and uses decision-level fusion to combine and analyze the results of different data extractions. Based on these two considerations, the model's operational efficiency is optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings of the embodiments of the present invention. The drawings are only used to illustrate some embodiments of the present invention, but not to limit all embodiments of the present invention thereto.

[0046] Figure 1 It is a flow chart of an underground space entrance and exit identification method based on multi-source data fusion according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the accompanying drawings of specific embodiments of the present invention to clearly and completely describe the exemplary embodiments of the present invention. Unless otherwise defined, technical or scientific terms used in the present invention should be given the common meanings understood by people with ordinary skills in the relevant field.

[0048] like Figure 1 As shown, this embodiment discloses a method for identifying underground space entrances and exits based on multi-source data fusion, which can effectively extract the spatial location distribution of underground space entrances and exits within the image coverage area, meeting the needs of large-scale underground space monitoring and expansion analysis; it specifically includes the following three steps.

[0049] Step S1: Use a hybrid pixel decomposition algorithm to process the optical image data, remove vegetation information and enhance soil spectral characteristics, and calculate the reflectance difference of each pixel.

[0050] This step takes into account the mixed effect of vegetation and soil. Based on historical empirical spectra, the mixed pixel decomposition algorithm is used to decompose the optical image data into mixed pixels, thereby removing vegetation information and enhancing soil spectral characteristics. The soil spectral data after pixel mixed pixel decomposition is then used as a reference to extract abnormal information.

[0051] (1) Green vegetation coverage is a key factor affecting the accuracy of ground entrance and exit information extraction. Due to differences in vegetation growth conditions, types, and canopy structures, vegetation spectra in different regions have different canopy characteristics, and their spectra have a certain range of variation. Here, the vegetation spectral range is determined based on historical empirical data and used as a constraint to constrain the mixed pixel decomposition results.

[0052] (2) Due to the large range of image data and the different soil background types, such as cement, sand, and asphalt, using the entire image data as the input parameter for mixed pixel decomposition will cause model errors. Here, the sliding window size is determined based on the number of mixed pixel bands in the image data, ensuring that the number of pixels in the window is much larger than the number of bands. For example, for data with a spatial resolution of 10m and 4 bands, the sliding window size is determined to be 20*20.

[0053] (3) Use the hybrid pixel decomposition algorithm to extract the soil reflectance of all pixels within each window

[0054] By eliminating the interference of vegetation spectra, the abnormal characteristics of soil reflectivity in the entrance and exit areas of underground spaces are enhanced. Specifically, because vegetation coverage will mask the true reflectivity of the soil (for example, the spectrum of densely vegetated areas is dominated by leaf reflection), and the soil around the entrances and exits of underground spaces often exhibits different reflectivity characteristics from natural soil due to human activities (such as compaction, material differences) or structural changes (such as exposure, humidity differences). By stripping off the vegetation signal through mixed pixel decomposition and extracting pure soil endmember spectra, the reflectivity difference between the soil in the entrance and exit area and the surrounding background can be directly quantified, thereby significantly improving the sensitivity and accuracy of anomaly detection. The use of mixed pixel decomposition algorithm to eliminate vegetation information solves the problems of false detection and missed detection caused by vegetation mixing effects in traditional methods.

[0055] Linear mixed pixel assumes that the mixed spectrum of vegetation and soil in a pixel of image data is the sum of the product of their respective area ratios and the corresponding end member spectra, that is, the model is expressed as:

[0056]

[0057] Among them, R leaf and R soil are vegetation and soil reflectance respectively, α and β are proportional coefficients, and ε is the error value.

[0058] ① Based on this principle, 400 mixed spectra in the sliding window are selected and 400 equations are constructed based on them. At the same time, 4 constraint equations are established based on the maximum / minimum values of soil reflectance and vegetation reflectance. Since each band corresponds to 404 equations, 1616 equations can be constructed for 4 bands. The linear decomposition of 1616 equations is realized by the least squares decomposition algorithm to extract α, β, R leaf and R soil .

[0059] ② The area ratio data α and β are smoothed and normalized using the mean smoothing method, and the spectrum of each pixel is recalculated based on the smoothed area ratio.

[0060] ③ Extract the soil end-member reflectance of all pixels in the sliding window, calculate the mean of the soil reflectance of all pixels in the sliding window, and calculate the difference between the soil reflectance of each pixel and the mean reflectance and perform normalization. The reflectance difference calculation formula is:

[0061]

[0062] Among them, B i is the soil reflectance of the current pixel in the i-th band, is the mean soil reflectance of all pixels in the sliding window, and M is the maximum difference value in the sliding window.

[0063] ④ Move the sliding window position and repeat the above steps until all image data are processed to generate a spatial distribution map of reflectivity differences of the entire image.

[0064] Step S2: Process the thermal infrared data and calculate the temperature difference of each pixel.

[0065] This step is based on thermal infrared data and extracts the location information of potential underground space entrances and exits based on historical experience data, such as the temperature differences of vegetation, ground, manhole covers or holes at different time periods.

[0066] The temperature difference with other surrounding landforms is also a feature for identifying underground space entrances and exits, including:

[0067] (1) The thermal infrared data are processed by atmospheric correction and radiation calibration, the original radiation values are converted into surface temperature values, and the data are noise filtered using the mean smoothing method.

[0068] (2) The thermal infrared data are processed using the Kriging spatial interpolation method to ensure that they have the same spatial resolution as the optical image data.

[0069] (3) Divide the data into blocks based on the sliding window (20*20) and calculate the temperature difference block by block, and perform normalization. The calculation formula is:

[0070]

[0071] Among them, H i is the temperature value of the current pixel, is the average temperature value of all pixels in the sliding window, and H is the maximum temperature difference value in the sliding window.

[0072] (4) Move the sliding window position and repeat the above steps until all image data are processed to generate a spatial distribution map of temperature differences for the entire image.

[0073] Step S3: Linearly fuse the reflectivity difference and temperature difference of each pixel to generate a pixel-level fusion result, and finally generate a spatial distribution map of the entrances and exits of the underground space.

[0074] Based on the linear mixing theory, the results obtained from two different data are fused at the decision level in a decision fusion manner to generate the final spatial distribution map of outliers. The outlier pixels / positions in each area are gradually determined by traversing the image with this window according to the image resolution and the window size.

[0075] (1) Based on the reflectivity difference of each pixel obtained in step S1 and the temperature difference of each pixel obtained in step S2, a linear fusion model is used to achieve pixel-level fusion of the two results. The fusion equation is:

[0076] D result =xD opt +yD the

[0077] Among them, D opt is the soil reflectivity, D the is the temperature difference, x and y are the decision fusion parameters, and x + y = 1. The values of the decision fusion parameters x and y are dynamically adjusted based on the temperature difference. The larger the temperature difference, the higher the reliability of the temperature signal, and the temperature weight y is increased, while the spectral weight x is decreased. Preferably, x = 0.8 and y = 0.2.

[0078] (2) Based on the sliding window range, the data within the range is differentiated and all extreme points within the range are extracted, i.e., the spatial locations of the entrances and exits of the underground space.

[0079] (3) Move the position of the sliding window and repeat the above steps until all image data are processed, and extract the spatial locations of the entrances and exits of the underground space in the image data.

[0080] Corresponding to the above method, this embodiment also discloses an underground space entrance and exit identification system based on multi-source data fusion, comprising:

[0081] The reflectance difference calculation module is used to process the optical image data using a hybrid pixel decomposition algorithm, remove vegetation information and enhance soil spectral characteristics, and calculate the reflectance difference of each pixel.

[0082] The temperature difference calculation module is used to process thermal infrared data and calculate the temperature difference of each pixel.

[0083] The multi-source data fusion module is used to linearly fuse the reflectivity difference and temperature difference of each pixel to generate pixel-level fusion results, and finally generate a spatial distribution map of the entrances and exits of the underground space.

[0084] In order to improve the accuracy of identifying underground space entrances and exits, the present invention is triggered from the perspectives of multi-source data fusion and abnormal feature enhancement, and constructs an underground space entrance and exit identification method based on the fusion of hybrid pixel decomposition signals and temperature signal anomaly detection. This method simultaneously considers both empirical knowledge and comparison of surrounding data of the same period images, which can effectively ensure the stability and accuracy of anomaly point extraction from image data.

[0085] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A method for identifying underground space entrances and exits based on multi-source data fusion, characterized in that: The following steps are involved: Step 1: Use the hybrid pixel decomposition algorithm to process the optical image data, remove vegetation information and enhance soil spectral characteristics, and calculate the reflectance difference of each pixel; Step 2: Process the thermal infrared data and calculate the temperature difference of each pixel; Step 3: Linearly fuse the reflectivity difference and temperature difference of each pixel to generate a pixel-level fusion result, and finally generate a spatial distribution map of the entrances and exits of the underground space.

2. The underground space entrance and exit identification method based on multi-source data fusion according to claim 1 is characterized in that: Step 1 specifically includes: Determine the vegetation spectral range based on historical empirical data and use it as a constraint to constrain the mixed pixel decomposition results; The sliding window size is determined according to the number of mixed pixel bands of the image data, and the image data is processed in blocks; The soil end-member reflectance of all pixels in each window is extracted using a mixed pixel decomposition algorithm, and the difference between the soil reflectance of each pixel and the window mean soil reflectance is calculated. The window mean soil reflectance refers to the average value of the soil end-member reflectance extracted from all pixels in the sliding window after mixed pixel decomposition.

3. The underground space entrance and exit identification method based on multi-source data fusion according to claim 2 is characterized in that: The mixed pixel decomposition algorithm is used to extract the soil endmember reflectance of all pixels in each window, and the difference between the soil reflectance of each pixel and the window soil mean reflectance is calculated, specifically including: In mixed pixel decomposition, the spectrum of each pixel is expressed as: Among them, R leaf and R soil are vegetation and soil reflectance, α and β are proportional coefficients, and ε is the error value; There are N pixels in each sliding window, and a total of M×(N+4) equations are generated, where M is the number of bands, and 4 is the four constraint equations established based on the lower and upper limits of soil reflectance and vegetation reflectance in each band. α, β, and R are solved by the least squares method. leaf and R soil ; Output the soil end member reflectance R of all pixels in the sliding window soil , calculate the mean soil reflectance of all pixels in the sliding window Calculate the soil reflectance R for each pixel soil and the mean reflectivity of the window soil difference.

4. The underground space entrance and exit identification method based on multi-source data fusion according to claim 3 is characterized in that: The formula for calculating the reflectivity difference is: Among them, B i is the soil reflectance of the current pixel in the i-th band, is the mean soil reflectance of all pixels in the sliding window, and M is the maximum difference value in the sliding window.

5. The underground space entrance and exit identification method based on multi-source data fusion according to claim 1 is characterized in that: Step 2 specifically includes: Perform atmospheric correction and radiation calibration on thermal infrared data and convert them into surface temperature values; The spatial resolution of thermal infrared data was adjusted to be consistent with that of optical image data using Kriging spatial interpolation method; The data is divided into blocks based on a sliding window and the temperature difference is calculated block by block.

6. The underground space entrance and exit identification method based on multi-source data fusion according to claim 5 is characterized in that: The temperature difference is calculated as: Among them, H i is the temperature value of the current pixel, is the average temperature value of all pixels in the sliding window, and H is the maximum temperature difference value in the sliding window.

7. The underground space entrance and exit identification method based on multi-source data fusion according to claim 1 is characterized in that: In step 3, the reflectivity difference and temperature difference of each pixel are linearly fused to generate a pixel-level fusion result, which is expressed as: D result =xD opt +yD the Among them, D opt is the soil reflectivity, D the is the temperature difference, x and y are decision fusion parameters, and x+y=1.

8. The underground space entrance and exit identification method based on multi-source data fusion according to claim 7 is characterized in that: The values of decision fusion parameters x and y are dynamically adjusted according to the temperature difference.

9. The underground space entrance and exit identification method based on multi-source data fusion according to claim 7 is characterized in that: The spatial distribution map of underground space entrances and exits generated in step 3 includes: The extreme points of the fused data are extracted based on the sliding window to determine the specific spatial locations of the entrances and exits of the underground space; Move the position of the sliding window and repeat the above steps until all image data are processed, and extract the spatial locations of the entrances and exits of the underground space in the image data.

10. An underground space entrance and exit identification system based on multi-source data fusion, characterized in that: A method for identifying underground space entrances and exits based on multi-source data fusion according to any one of claims 1 to 9, comprising: The reflectance difference calculation module is used to process the optical image data using a hybrid pixel decomposition algorithm, remove vegetation information and enhance soil spectral characteristics, and calculate the reflectance difference of each pixel; Temperature difference calculation module, used to process thermal infrared data and calculate the temperature difference of each pixel; The multi-source data fusion module is used to linearly fuse the reflectivity difference and temperature difference of each pixel to generate pixel-level fusion results, and finally generate a spatial distribution map of the entrances and exits of the underground space.

Images