Methane concentration spatial distribution inversion method and system

Abstract

This invention discloses a method and system for inverting the spatial distribution of methane concentration. The method includes: acquiring hyperspectral images within the characteristic absorption band range of methane to construct a data cube; performing Gaussian smoothing and noise reduction on the data cube pixel by pixel to obtain a smoothed data cube; selecting the optimal fitting order for the data in each band of the smoothed data cube and performing polynomial fitting to obtain baseline spectral values; calculating absorbance based on the denoised and smoothed spectral intensity and the baseline spectral values, and setting physical constraints; solving for the integral width and imposing upper and lower limits on the integral width; performing integral calculation on the absorbance curve to obtain the absorption peak area; performing maximum and minimum normalization on the absorption peak area to output the relative methane concentration. The advantage of this invention is that it improves the accuracy and reliability of quantitative inversion of methane concentration.

Description

Technical Field

[0001] This invention relates to the field of gas detection technology, specifically to a method and system for inverting the spatial distribution of methane concentration. Background Technology

[0002] Methane is a major component of natural gas and a key component of coal mine gas. During natural gas pipeline transportation and storage, factors such as equipment aging and deterioration of sealing structures can easily lead to methane leaks. Leaked methane not only wastes energy resources and, as a potent greenhouse gas, severely impacts the atmospheric environment, but also, when methane accumulates in confined spaces to its explosive limits, it can cause serious safety accidents such as fires or explosions. Furthermore, during coal mining, methane accumulation caused by gas outbursts is a primary threat to mine safety, posing a continuous danger to miners' lives. Therefore, achieving rapid and accurate detection and leak source location of methane leaks is of significant practical importance for ensuring the safe operation of energy facilities and safe coal mine production.

[0003] Tunable Diode Laser Absorption Spectroscopy (TDLAS) utilizes a diode laser to scan the wavelength of a target gas within its characteristic absorption band, using the change in light intensity before and after absorption to characterize gas concentration. This technique effectively reduces interference from other gas components by selecting relatively independent absorption lines for the target gas, ensuring the accuracy of measurement results and offering advantages such as high sensitivity, high selectivity, and rapid response. By combining wavelength modulation and harmonic detection techniques, high-precision and rapid measurement of trace gases can be achieved even at low absorption intensities. In recent years, with advancements in the manufacturing process of area array detectors and improved data processing capabilities, combining TDLAS technology with hyperspectral imaging to construct a novel "image-spectrum integration" detection method has become an important development direction in the field of gas detection. This method can simultaneously acquire spatial and spectral information of the target area, offering significant advantages over traditional single-point measurement methods in visualizing gas concentration distribution and accurately locating leak sources. It shows broad application prospects in scenarios such as natural gas station leak monitoring, pipeline inspection, and coal mine safety. However, existing concentration inversion methods based on hyperspectral imaging still suffer from insufficient accuracy in key stages of spectral data processing, which limits the accuracy and reliability of quantitative methane concentration inversion. Summary of the Invention

[0004] The technical problem to be solved by this invention is how to improve the accuracy and reliability of quantitative inversion of methane concentration.

[0005] This invention solves the above-mentioned technical problems through the following technical means: a method for inverting the spatial distribution of methane concentration, comprising the following steps: S1. Collect hyperspectral images within the characteristic absorption band of methane to construct a data cube; S2. Perform Gaussian smoothing on the data cube pixel by pixel to reduce noise, resulting in a smoothed data cube; S3. Select the optimal fitting order for the data of each band in the smoothed data cube, perform polynomial fitting, and obtain the baseline spectral values. S4. Calculate the absorbance based on the denoised and smoothed spectral intensity and the baseline spectral value, and set physical constraints. S5. Solve for the integral width and impose upper and lower limits on the integral width; S6. Determine the integration interval based on the integration width, and calculate the absorption peak area by integrating the absorbance curve within the integration interval. S7. Normalize the absorption peak area to the maximum and minimum to output the relative concentration of methane.

[0006] Further, S1 includes: n frames of single-channel grayscale spectral images were acquired at equal intervals within the characteristic absorption band of methane. All band images were then stacked along the spectral dimension in ascending order of wavelength, constructing a dimension-... The data cube, in which, The height is the number of pixels in the image. The image width in pixels. The total number of bands represents any pixel position within the data cube. This corresponds to a one-dimensional spectral vector containing information for all spectral bands.

[0007] Further, S3 includes: Using the center band index of the methane characteristic absorption peak as a benchmark, stable bands without gas absorption features on both sides are selected as the baseline fitting interval. For each pixel in the image, the band index and its corresponding smoothed spectral intensity within the baseline fitting interval are extracted to construct a baseline fitting dataset. Least square fitting is performed on the same baseline fitting dataset using first-order and second-order polynomials, respectively. The mean square error corresponding to each of the two fitting orders is calculated, and the polynomial with the smaller mean square error is selected as the optimal baseline model for the current pixel. Based on the polynomial coefficients of the optimal baseline model, interpolation is used to calculate the baseline spectral values ​​corresponding to all bands. , For the first Each band.

[0008] Furthermore, the formula for calculating the absorbance is as follows:

[0009] in, The intensity is the spectral intensity after noise reduction and smoothing.

[0010] Furthermore, the physical constraint condition is that when the spectral intensity ratio Or, if the calculated absorbance value is negative, force the band to be... The absorbance is set to 0.

[0011] Further, S5 includes: Extract the maximum absorbance value at the center of the absorption peak. ,by Using this as a baseline, a bidirectional search was performed to the left and right of the spectrum to locate the absorbance value at which it decayed. The band indices corresponding to the left and right boundaries are used, and the difference between the band indices of the right and left boundaries is taken as the actual full width at half maximum (FWHM) of the methane absorption peak. Set a fixed ratio coefficient ,pass The integral width is calculated, and upper and lower limits are imposed on the integral width, forcibly limiting the effective integral width to 3 to 8 band intervals.

[0012] Further, S6 includes: Using the center band of the methane absorption peak as a reference, the final integration interval is determined by combining the integration width. Within the final integration interval, the Simpson numerical integration algorithm is used to perform integration on the absorbance curve to obtain the area of ​​the methane characteristic absorption peak corresponding to the current pixel.

[0013] Further, S7 includes: S71. Calculate the area of ​​absorption peaks of all pixels in the global domain, truncate outliers, and obtain the truncated area at each pixel position. S72. Map the area of ​​each pixel location after truncation to the relative concentration range of 0 to 100 by using minimum-maximum normalization to obtain the relative concentration of methane. S73. Generate a heat map of the spatial distribution of methane concentration based on the global pixel concentration matrix.

[0014] Furthermore, S71 includes: Calculate the 1st percentile of the global pixel absorption peak area. 99th percentile of global pixel absorption peak area Truncate outliers to obtain pixel positions. Area after truncation:

[0015] in, pixel position The area of ​​the absorption peak at that point.

[0016] The present invention also provides a methane concentration spatial distribution inversion system, which performs the above-described method, including: The data acquisition module is used to acquire hyperspectral images within the characteristic absorption band range of methane to construct a data cube; The data smoothing module is used to perform pixel-by-pixel Gaussian smoothing and noise reduction on the data cube to obtain a smoothed data cube. The polynomial fitting module is used to select the optimal fitting order for the data of each band in the smoothed data cube, perform polynomial fitting, and obtain the baseline spectral values. The absorbance calculation module is used to calculate absorbance based on the noise-reduced and smoothed spectral intensity and the baseline spectral value, and to set physical constraints. The integral width calculation module is used to calculate the integral width and impose upper and lower limits on the integral width. The absorption peak area calculation module is used to determine the integration interval based on the integration width, and to perform integration calculation on the absorbance curve within the integration interval to obtain the absorption peak area. The concentration inversion module is used to perform maximum and minimum normalization on the absorption peak area and output the relative concentration of methane.

[0017] The advantages of this invention are: (1) The present invention uses Gaussian smoothing filter to preprocess the spectrum of each pixel, effectively suppressing noise while maintaining the original shape of the absorption peak, avoiding peak attenuation and peak distortion caused by traditional filtering; adopts an adaptive order selection strategy, automatically selects the optimal fitting order from the first-order polynomial and the second-order polynomial according to the actual spectral shape of each pixel, fundamentally overcoming the contradiction that traditional fixed-order fitting cannot simultaneously meet the fitting requirements of straight lines and curved baselines, significantly improving the concentration inversion accuracy, and improving the accuracy and reliability of quantitative inversion of methane concentration.

[0018] (2) This invention uses Simpson's integral method instead of the traditional trapezoidal integral method, which effectively improves the integration accuracy; at the same time, it adaptively determines the integration width based on the actual half-width of each pixel, avoiding the peak area calculation deviation caused by fixed-width integration. This invention independently completes the entire process for each pixel in the data cube, directly generating a concentration distribution image, realizing the integrated inversion of accurate location of methane leakage and quantitative concentration analysis. Attached Figure Description

[0019] Figure 1 This is a flowchart of a method for inverting the spatial distribution of methane concentration disclosed in an embodiment of the present invention; Figure 2 This is a schematic diagram of the experimental system structure in a method for inverting the spatial distribution of methane concentration disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of the methane absorption wavelength in a methane concentration spatial distribution inversion method disclosed in an embodiment of the present invention; Figure 4This is a schematic diagram illustrating the automatic identification of absorption peak positions using the center band index of methane characteristic absorption peaks in a methane concentration spatial distribution inversion method disclosed in an embodiment of the present invention. Figure 5(a) is a schematic diagram of the original spectrum of the background region and the fitted baseline in a methane concentration spatial distribution inversion method disclosed in the embodiment of the present invention, and Figure 5(b) is the absorbance curve corresponding to Figure 5(a); Figure 6 In the methane concentration spatial distribution inversion method disclosed in the embodiments of the present invention, 10% 4 and 5% 4. Intensity 2D Concentration Distribution Heat Map of Airbags; Figure 7 In the methane concentration spatial distribution inversion method disclosed in the embodiments of the present invention, 10% 4 and 5% 4. 3D concentration distribution surface diagram of airbag strength. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1 like Figure 1 and Figure 2 As shown, Embodiment 1 of the present invention provides a method for inverting the spatial distribution of methane concentration. Through techniques such as Gaussian smoothing and noise reduction, adaptive polynomial baseline fitting, adaptive integral width determination, and Simpson integral, a high-precision spatial distribution inversion of methane gas concentration is achieved. The method includes the following steps: S1. Collect hyperspectral images within the characteristic absorption band of methane to construct a data cube; Continue reading Figure 2The data processing and control unit sends control commands to the laser driver, which then drives the tunable semiconductor laser to generate laser light of a specific wavelength, with the center wavelength aligned with the characteristic absorption peak of methane molecules. The original beam emitted from the tunable semiconductor laser is collimated and homogenized by an optical shaping module to eliminate spatial inhomogeneities and form a uniformly distributed divergent beam, which is then projected onto the target gas region. After the laser interacts with the methane molecules in the target gas region, the emitted light carrying gas absorption information is collected by an imaging lens and focused onto the focal plane of a hyperspectral detector. The hyperspectral detector converts the light signal into a digital image signal, and finally returns the acquired multi-band spectral image to the data processing and control unit for storage and subsequent concentration inversion processing.

[0022] This method selects the characteristic absorption band range of methane. Fifty single-channel grayscale spectral images were acquired at equal intervals within this band. All band images were then stacked along the spectral dimension in ascending order of wavelength, constructing a dimensional array. A three-dimensional hyperspectral data cube, in which, The height is the number of pixels in the image. 50 represents the image width in pixels, and 50 represents the total number of bands. (Any pixel position within the data cube) This corresponds to a one-dimensional spectral vector containing information from all 50 bands, fully preserving the original information in both spatial and spectral dimensions, thus providing a data foundation for subsequent pixel-by-pixel refined inversion.

[0023] S2. Perform Gaussian smoothing on the data cube pixel by pixel to reduce noise, resulting in a smoothed data cube; Gaussian smoothing filtering is applied to the one-dimensional spectral vectors of each pixel in the data cube to suppress random spectral noise and curve jitter. The filtering parameter is fixed at the Gaussian standard deviation. Filter window radius The smoothing formula is:

[0024] In the formula, This is the index for the current band. For the current band index The window offset (t=-m,-m+1,...,m). For the first Original spectral intensity values ​​of the band This represents the filtered spectral intensity value. These are the Gaussian kernel weighting coefficients. This step achieves smoothing preprocessing of the spectral curve, suppressing high-frequency noise while preserving characteristic absorption peaks.

[0025] S3. Select the optimal fitting order for the data of each band in the smoothed data cube, perform polynomial fitting, and obtain the baseline spectral values. like Figure 3 As shown, indexed by the center band of the characteristic absorption peak of methane. Based on the baseline, avoiding the effective signal region of the absorption peak, the stable bands with no gas absorption characteristics on both sides are selected as the baseline fitting interval, specifically the left band index 1 to 20 and the right band index 31 to 50. Figure 3 Indicates the absorption wavelength of the methane center. and characteristic absorption band range ~ This is used to illustrate that the spectral detection range selected in this invention covers the characteristic absorption peaks of methane molecules, providing a wavelength basis for subsequent hyperspectral data acquisition.

[0026] like Figure 4 As shown, for each pixel in the image, a baseline fitting dataset is constructed with the band index as the independent variable x and the filtered spectral intensity value as the dependent variable y. Least square fitting is performed on the same baseline fitting dataset using first-order and second-order polynomials, respectively, fitting the curves showing the relationship between spectral intensity and band variation. The mean square error corresponding to each of the two fitting orders is calculated. The optimal fitting order is adaptively selected through error comparison: the polynomial with the smaller mean square error is chosen as the optimal baseline model for the current pixel. Based on the polynomial coefficients of the optimal baseline model, interpolation is used to calculate the baseline spectral values ​​corresponding to all 50 bands. That is, the first bands The corresponding spectral intensity value. Figure 4 The image shows the average original spectral curve of the background region and the spectral curve after Gaussian smoothing. The smoothed curve retains the absorption peak characteristics while effectively suppressing spectral noise. The trough position corresponds to the center of the methane characteristic absorption peak. The algorithm automatically locates the peak index p=25, which is used to verify the consistency and repeatability of the automatic location of absorption peak positions in spectral data in this invention.

[0027] S4. Calculate the absorbance and set physical constraints; As shown in Figure 5, based on the Lambert-Beer absorption law, the absorbance value of the corresponding spectrum of a pixel is calculated band by band using the smoothed spectral intensity and the baseline spectral intensity obtained by adaptive fitting. The calculation formula is as follows:

[0028] in, The intensity is the spectral intensity after noise reduction and smoothing.

[0029] To eliminate noise outliers and non-physical anomalies caused by uneven illumination, strict physical constraints are set: when the spectral intensity ratio If the calculated absorbance value is negative, the absorbance of that band will be forcibly set to 0.

[0030] Figure 5(a) shows the band index on the horizontal axis and the image grayscale intensity on the vertical axis, displaying the smoothed spectral curve (blue solid line) and the adaptively fitted baseline curve (red dashed line), which is used to verify that the adaptive baseline fitting method of the present invention can accurately track the overall trend of the spectrum in the non-absorption region; Figure 5(b) shows the absorbance curve calculated based on the spectrum and baseline of Figure 5(a), which is used to verify that the absorbance tends to zero baseline outside the absorption peak position, indicating the accuracy of the baseline fitting and the rationality of the absorbance calculation.

[0031] S5. Extract the maximum absorbance value at the center position of the methane absorption peak. Using this peak value as a reference, a bidirectional search is performed to the left and right of the spectrum to locate the absorbance value that has decreased to [value missing]. The corresponding left and right boundary band indices L and R are used, and the difference (RL) between the left and right boundary band indices is taken as the actual full width at half maximum (FWHM) of the methane absorption peak. Set a fixed proportional coefficient. ,pass The integration width is calculated. To avoid introducing background noise due to an excessively large integration interval or losing weak absorption features due to an excessively small interval, upper and lower limits are imposed on the integration width, forcibly limiting the effective integration width to 3 to 8 band intervals.

[0032] S6. Using the center band of the methane absorption peak as a reference, and combining it with the obtained integral width, determine the integration interval. Within this interval, the Simpson numerical integration algorithm is used to perform high-precision integration on the absorbance curve, solving for the total area of ​​the methane characteristic absorption peak corresponding to the current pixel. The absorption peak area accurately characterizes the overall light absorption of methane gas at the pixel location and is a core feature parameter for correlating methane concentration. Compared to traditional trapezoidal integration, Simpson integration can accurately fit a smooth spectral absorption curve, significantly improving the extraction accuracy and stability of weak methane absorption signals.

[0033] S7. Concentration Normalization and Spatial Distribution Inversion 1) Quantile outlier truncation: Calculate the 1% quantile of the absorption peak area of ​​all pixels in the global pixel area. 99th percentile of global pixel absorption peak area By truncating outliers, we can obtain the truncated area at each pixel location:

[0034] in, Indicates pixel position The area of ​​the absorption peak at that point.

[0035] 2) Min-Max normalization for relative concentration: Map the truncated area to the 0-100 relative concentration range:

[0036] In the formula, For pixels The corresponding relative concentration of methane, This represents the area after global pixel truncation.

[0037] 3) Visual imaging: such as Figure 6 and Figure 7 As shown, based on the global pixel density matrix ( The first pixel in the global pixel density matrix Line 1 (The data is analyzed using a series of elements) to generate a thermal map of the spatial distribution of methane concentration, enabling the location of the leak area and the visualization and inversion of the concentration gradient. Figure 6 The horizontal and vertical axes represent pixel positions, and the color intensity indicates the relative concentration. This verifies that the present invention can achieve visualization imaging of the spatial distribution of methane concentration and can clearly distinguish the spatial position and concentration differences of gas bags with different concentrations. Figure 7 The XY plane represents the pixel position, and the Z-axis height represents the relative concentration level. This is used to visually verify that the concentration distribution inverted by this invention exhibits a bimodal distribution characteristic in space. The concentration peaks of the two gas bags are consistent with the actual concentration ratio, further verifying the effectiveness of this method in locating the leakage source and quantitatively analyzing the concentration.

[0038] Through the above technical solutions, this invention employs Gaussian smoothing filtering to preprocess the spectrum of each pixel, effectively suppressing noise while maintaining the original shape of the absorption peaks, avoiding peak attenuation and peak distortion caused by traditional filtering. It adopts an adaptive order selection strategy, automatically selecting the optimal fitting order from first-order and second-order polynomials based on the actual spectral shape of each pixel, fundamentally overcoming the contradiction that traditional fixed-order fitting cannot simultaneously meet the fitting requirements of straight and curved baselines, significantly improving the concentration inversion accuracy. It uses the Simpson integral method instead of the traditional trapezoidal integral method, effectively improving integration accuracy; simultaneously, it adaptively determines the integration width based on the actual half-width of each pixel, avoiding peak area calculation deviations caused by fixed-width integration. This invention independently completes the entire process for each pixel in the data cube, directly generating a concentration distribution image, realizing the integrated inversion of accurate methane leakage location and quantitative concentration analysis.

[0039] Example 2 Embodiment 2 of the present invention provides a methane concentration spatial distribution inversion system, which executes the method of Embodiment 1, including: The data acquisition module is used to acquire hyperspectral images within the characteristic absorption band range of methane to construct a data cube; The data smoothing module is used to perform pixel-by-pixel Gaussian smoothing and noise reduction on the data cube to obtain a smoothed data cube. The polynomial fitting module is used to select the optimal fitting order for the data of each band in the smoothed data cube, perform polynomial fitting, and obtain the baseline spectral values. The absorbance calculation module is used to calculate absorbance based on the noise-reduced and smoothed spectral intensity and the baseline spectral value, and to set physical constraints. The integral width calculation module is used to calculate the integral width and impose upper and lower limits on the integral width. The absorption peak area calculation module is used to determine the integration interval based on the integration width, and to perform integration calculation on the absorbance curve within the integration interval to obtain the absorption peak area. The concentration inversion module is used to perform maximum and minimum normalization on the absorption peak area and output the relative concentration of methane.

[0040] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for inverting the spatial distribution of methane concentration, characterized in that, Includes the following steps: S1. Collect hyperspectral images within the characteristic absorption band of methane to construct a data cube; S2. Perform Gaussian smoothing on the data cube pixel by pixel to reduce noise, resulting in a smoothed data cube; S3. Select the optimal fitting order for the data of each band in the smoothed data cube, perform polynomial fitting, and obtain the baseline spectral values. S4. Calculate the absorbance based on the denoised and smoothed spectral intensity and the baseline spectral value, and set physical constraints. S5. Solve for the integral width and impose upper and lower limits on the integral width; S6. Determine the integration interval based on the integration width, and calculate the absorption peak area by integrating the absorbance curve within the integration interval. S7. Normalize the absorption peak area to the maximum and minimum to output the relative concentration of methane.

2. The method for inverting the spatial distribution of methane concentration according to claim 1, characterized in that, S1 includes: n frames of single-channel grayscale spectral images were acquired at equal intervals within the characteristic absorption band of methane. All band images were then stacked along the spectral dimension in ascending order of wavelength, constructing a dimension-... The data cube, in which, The height is the number of pixels in the image. The image width in pixels. The total number of bands represents any pixel position within the data cube. This corresponds to a one-dimensional spectral vector containing information for all spectral bands.

3. The method for inverting the spatial distribution of methane concentration according to claim 1, characterized in that, S3 includes: Using the center band index of the methane characteristic absorption peak as a benchmark, stable bands without gas absorption features on both sides are selected as the baseline fitting interval. For each pixel in the image, the band index and its corresponding smoothed spectral intensity within the baseline fitting interval are extracted to construct a baseline fitting dataset. Least square fitting is performed on the same baseline fitting dataset using first-order and second-order polynomials, respectively. The mean square error corresponding to each of the two fitting orders is calculated, and the polynomial with the smaller mean square error is selected as the optimal baseline model for the current pixel. Based on the polynomial coefficients of the optimal baseline model, interpolation is used to calculate the baseline spectral values ​​corresponding to all bands. , For the first Each band.

4. The method for inverting the spatial distribution of methane concentration according to claim 3, characterized in that, The formula for calculating the absorbance is as follows: in, The intensity is the spectral intensity after noise reduction and smoothing.

5. The method for inverting the spatial distribution of methane concentration according to claim 4, characterized in that, The physical constraint condition is that when the spectral intensity ratio Or, if the calculated absorbance value is negative, force the band to be... The absorbance is set to 0.

6. The method for inverting the spatial distribution of methane concentration according to claim 1, characterized in that, S5 includes: Extract the maximum absorbance value at the center of the absorption peak. ,by Using this as a baseline, a bidirectional search was performed to the left and right of the spectrum to locate the absorbance value at which it decayed. The band indices corresponding to the left and right boundaries are used, and the difference between the band indices of the right and left boundaries is taken as the actual full width at half maximum (FWHM) of the methane absorption peak. Set a fixed ratio coefficient ,pass The integral width is calculated, and upper and lower limits are imposed on the integral width, limiting the effective integral width to 3 to 8 band intervals.

7. The method for inverting the spatial distribution of methane concentration according to claim 1, characterized in that, S6 includes: Using the center band of the methane absorption peak as a reference, the integration interval is determined by combining the integration width. Within the integration interval, the Simpson numerical integration algorithm is used to perform integration on the absorbance curve to obtain the area of ​​the methane characteristic absorption peak corresponding to the current pixel.

8. The method for inverting the spatial distribution of methane concentration according to claim 1, characterized in that, S7 includes: S71. Calculate the area of ​​absorption peaks of all pixels in the global domain, truncate outliers, and obtain the truncated area at each pixel position. S72. Map the area of ​​each pixel location after truncation to the relative concentration range of 0 to 100 by using minimum-maximum normalization to obtain the relative concentration of methane. S73. Generate a heat map of the spatial distribution of methane concentration based on the global pixel concentration matrix.

9. The method for inverting the spatial distribution of methane concentration according to claim 8, characterized in that, S71 includes: Calculate the 1st percentile of the global pixel absorption peak area. 99th percentile of global pixel absorption peak area Truncate outliers to obtain pixel positions. Area after truncation: in, pixel position The area of ​​the absorption peak at that point.

10. A methane concentration spatial distribution inversion system, characterized in that, The method of any one of claims 1-9 comprises: The data acquisition module is used to acquire hyperspectral images within the characteristic absorption band range of methane to construct a data cube; The data smoothing module is used to perform pixel-by-pixel Gaussian smoothing and noise reduction on the data cube to obtain a smoothed data cube. The polynomial fitting module is used to select the optimal fitting order for the data of each band in the smoothed data cube, perform polynomial fitting, and obtain the baseline spectral values. The absorbance calculation module is used to calculate absorbance based on the noise-reduced and smoothed spectral intensity and the baseline spectral value, and to set physical constraints. The integral width calculation module is used to calculate the integral width and impose upper and lower limits on the integral width. The absorption peak area calculation module is used to determine the integration interval based on the integration width, and to perform integration calculation on the absorbance curve within the integration interval to obtain the absorption peak area. The concentration inversion module is used to perform maximum and minimum normalization on the absorption peak area and output the relative concentration of methane.

Images