A method for removing the influence of moisture on hyperspectral reflectance data

Abstract

The present invention provides a method for removing the influence of moisture on hyperspectral reflectance data, which relates to the field of hyperspectral remote sensing technology. The method includes: performing spectral denoising preprocessing on the acquired original hyperspectral reflectance data of the ground object in its natural state according to the noise level to obtain target hyperspectral reflectance data; calculating a moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data; using the matrix composed of the moisture factors as a reference matrix, removing the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain corrected spectral data, and performing a correlation calculation between the corrected spectral data and a standard spectral library to obtain the final corrected spectrum after removing the influence of moisture. This application can remove moisture from hyperspectral reflectance data on the basis of maintaining the inherent characteristics of the ground object spectrum and without relying on external measured auxiliary data, improving the applicability of subsequent spectral application models.

Description

Technical Field

[0001] The present invention relates to the field of hyperspectral remote sensing technology, and in particular to a method for removing the influence of moisture on hyperspectral reflectance data. Background Art

[0002] Existing research shows that moisture has a non-linear effect on hyperspectral reflectance data. Taking a damp mural as an example, when the moisture content is low, an increase in moisture will reduce the spectral reflectance; while when the moisture content is high, as the moisture further increases, the spectral reflectance will instead increase. Moisture not only changes the spectral reflectance value on the surface of the mural, but also affects the shape of the spectral curve, thereby interfering with the characteristic absorption peaks of substances such as mineral pigments and substrate components in the mural material, and affecting the accuracy of information extraction in subsequent spectral application models. Therefore, removing the influence of moisture on hyperspectral reflectance data is an important prerequisite for ensuring the reliability of spectral data, improving the applicability of subsequent spectral application models, and achieving accurate analysis and effective application.

[0003] Existing research mostly removes the influence of environmental factors such as moisture through spectral transformation methods. Among them, the DS and EPO algorithms are the most commonly used, but both require relying on laboratory spectra to complete environmental factor correction. The principle of the DS algorithm is to establish a conversion matrix between the spectrum of the target object without moisture (abbreviated as dry spectrum) and the spectrum of the target object with moisture (abbreviated as wet spectrum) through a standardization method, making the wet spectrum as similar as possible to the dry spectrum, so as to achieve the purpose of removing the influence of moisture. The principle of the EPO algorithm is to construct a conversion matrix through the difference matrix between the wet spectrum and the dry spectrum, separate the information related to moisture from the wet spectrum, and finally retain the information unrelated to moisture to achieve the purpose of removing the influence of moisture.

[0004] In the prior art, moisture influence removal algorithms such as DS and EPO all require relying on measured auxiliary data (laboratory spectra of dry target objects), which increases the complexity of data acquisition and processing. Moreover, it is difficult to obtain dry spectra in large-scale image applications, which limits the applicability of the method. In addition, the EPO algorithm will also change the spectral shape, and it is impossible to directly match and identify ground objects and extract ground object information through the spectra in the standard spectral library, thereby affecting the accuracy of target recognition and information extraction in subsequent spectral application models. Summary of the Invention

[0005] The present invention provides a method for removing the influence of moisture on hyperspectral reflectance data, which can remove moisture from hyperspectral reflectance data on the basis of maintaining the inherent characteristics of the ground object spectrum and without relying on external measured auxiliary data, and improves the applicability of subsequent spectral application models.

[0006] The present invention provides a method for removing the influence of moisture on hyperspectral reflectance data, and the method includes the following steps:

[0007] Obtain the original hyperspectral reflectance data of the ground objects in their natural state;

[0008] Perform spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain the target hyperspectral reflectance data;

[0009] Calculate the moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data;

[0010] Using the matrix composed of the moisture factors as the reference matrix, remove the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain the corrected spectral data, and perform a correlation calculation between the corrected spectral data and the standard spectral library to obtain the final corrected spectrum after removing the influence of moisture.

[0011] In some embodiments, the performing spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain the target hyperspectral reflectance data includes:

[0012] Remove the spectral data of the high-noise bands affected by water vapor in the original hyperspectral reflectance data to obtain the spectral data to be processed;

[0013] Divide the spectral data to be processed into spectral data of high-noise bands, low-noise bands, and medium-noise bands according to the signal-to-noise ratio;

[0014] Perform denoising processing on the spectral data of the high-noise bands, low-noise bands, and medium-noise bands respectively to obtain the target hyperspectral reflectance data;

[0015] Among them, the process of denoising processing includes:

[0016] For the spectral data to be processed in the high-noise bands, use the wavelet threshold denoising method for denoising;

[0017] For the spectral data to be processed in the low-noise bands, use the moving average algorithm for denoising;

[0018] For the spectral data to be processed in the medium-noise bands, use the Savitzky-Golay filtering algorithm for denoising.

[0019] In some embodiments, the calculating the moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data includes:

[0020] Determine the short-wave infrared bands that are sensitive to the moisture content;

[0021] For the target hyperspectral reflectance data, calculate the short-wave infrared moisture stress index, the normalized short-wave infrared moisture index, and the visible and short-wave infrared drying index respectively according to the short-wave infrared bands as the moisture factors.

[0022] In some embodiments, removing information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain corrected spectral data includes:

[0023] Extracting eigencomponents from the spectral data matrix corresponding to the target hyperspectral reflectance data by using the singular value decomposition algorithm;

[0024] Constructing a projection matrix of the spectral data matrix in the orthogonal direction of the reference matrix;

[0025] Performing eigenmapping of the eigencomponents in the projection matrix to obtain corrected spectral data.

[0026] In some embodiments, removing the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data is implemented by an adaptive iterative optimization strategy, and the adaptive iterative optimization strategy includes:

[0027] Presetting a convergence threshold for removing the influence of moisture;

[0028] If the current corrected spectral data does not reach the convergence threshold, then iteratively execute the following process:

[0029] Removing the information features related to the reference matrix from the current corrected spectral data;

[0030] If the current corrected spectral data reaches the convergence threshold, then end the iteration.

[0031] In some embodiments, calculating the correlation between the corrected spectral data and a standard spectral library to obtain the final corrected spectral data after removing the influence of moisture includes:

[0032] Calculating the spectral correlation coefficient between the corrected spectral data and the standard spectral library;

[0033] Selecting the corrected spectral data with the largest spectral correlation coefficient as the final corrected spectral data after removing the influence of moisture.

[0034] The present invention also provides a device for removing the influence of moisture from hyperspectral reflectance data, and the device includes the following modules:

[0035] An acquisition module, configured to acquire the original hyperspectral reflectance data of a ground object in its natural state;

[0036] A preprocessing module, configured to perform spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain target hyperspectral reflectance data;

[0037] A calculation module, configured to calculate a moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data;

[0038] A removal module, configured to use the matrix composed of the moisture factors as a reference matrix, remove information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain corrected spectral data, and perform a correlation calculation between the corrected spectral data and a standard spectral library to obtain a final corrected spectrum after removing the moisture influence.

[0039] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method for removing the moisture influence of the hyperspectral reflectance data as described in any one of the above is implemented.

[0040] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for removing the moisture influence of the hyperspectral reflectance data as described in any one of the above is implemented.

[0041] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, the method for removing the moisture influence of the hyperspectral reflectance data as described in any one of the above is implemented.

[0042] The method for removing the moisture influence of the hyperspectral reflectance data provided by the present invention directly uses the preprocessed original hyperspectral reflectance data, calculates and analyzes the moisture factor that can accurately characterize the water content of the target object based on the short-wave infrared band, without relying on external measured auxiliary data, greatly simplifying the data acquisition and processing process. Only by using the moisture factor calculated from the hyperspectral reflectance data as a reference can the moisture influence be removed, significantly reducing the dependence on external measured data, improving the applicability of the subsequent spectral application model, and at the same time being able to keep the inherent characteristics of the ground object spectrum unchanged, and can provide technical support for the accurate application of hyperspectral data in multiple fields such as multi-temporal, multi-scene, and large-scale quantitative remote sensing, information extraction, and target recognition. Description of the Drawings

[0043] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art one by one. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0044] Figure 1 It is a flowchart of the method for removing the moisture influence of the hyperspectral reflectance data provided by the present invention.

[0045] Figure 2 It is a schematic structural diagram of a device for removing the influence of moisture on hyperspectral reflectance data provided by the present invention.

[0046] Figure 3 It is a schematic structural diagram of an electronic device provided by the present invention. Specific embodiments

[0047] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0048] The method for removing the influence of moisture on hyperspectral reflectance data of the present invention will be described below with reference to the accompanying drawings. Figure 1 It is a schematic flowchart of the method for removing the influence of moisture on hyperspectral reflectance data provided by the present invention. As Figure 1 shown, the method includes the following steps 101 to 104, which will be specifically described below: Step 101, obtain the original hyperspectral reflectance data of the ground object in its natural state.

[0049] First, collect and obtain the original hyperspectral reflectance data of the ground object in its natural state. During the collection process, use a portable ground object spectrometer or imaging spectrometer to perform spectral measurement on the remote sensing ground object in its natural state, and then obtain the original hyperspectral reflectance data of the ground object in the wavelength range of 400 nm - 2500 nm. Among them, the natural state refers to the normal state without interference from other human factors, and the original hyperspectral reflectance data can be point spectral reflectance data or hyperspectral reflectance images.

[0050] Step 102, perform spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain the target hyperspectral reflectance data.

[0051] The obtained original hyperspectral reflectance data generally needs to be preprocessed. First, remove some bands severely affected by noise, such as bands severely affected by water vapor. In addition, in order to retain spectral information to the greatest extent, further denoising processing is also required. In the embodiments of the present invention, according to the noise level of the spectral data, different denoising algorithms can be used to perform spectral denoising preprocessing on the original hyperspectral reflectance data to obtain the target hyperspectral reflectance data. The denoising algorithms can be Savitzky-Golay filtering algorithm, wavelet threshold denoising method, moving average algorithm, etc.

[0052] Step 103: Calculate the moisture factor related to the shortwave infrared band based on the target hyperspectral reflectance data.

[0053] For the preprocessed target hyperspectral reflectance data, embodiments of the present invention directly calculate the moisture factor to accurately characterize the moisture content as much as possible. The shortwave infrared band with higher sensitivity to moisture content is preferentially selected. Based on the selected shortwave infrared band, multiple moisture factors are calculated, including but not limited to: Shortwave Infrared Water Stress Index (SIWSI), Normalized Shortwave Infrared Water Index (NSWI), and Visible and Shortwave Infrared Drought Index (VSDI). When calculating the above moisture factors, all band combinations within the involved wavelength range are traversed and calculated one by one. In addition, for different targets and application scenarios, the wavelength range used to calculate the moisture factor can be dynamically adjusted to effectively improve the practicability and stability of the moisture factor in different scenarios.

[0054] Step 104: Using the matrix composed of moisture factors as the reference matrix, remove the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain corrected spectral data, and calculate the correlation between the corrected spectral data and the standard spectral library to obtain the final corrected spectral after removing the influence of moisture.

[0055] After calculating the moisture factor of the spectral data, use the matrix composed of moisture factors as the reference matrix. Then, remove the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain corrected spectral data, thereby eliminating the moisture features in the target hyperspectral reflectance data and achieving the purpose of removing the influence of moisture.

[0056] The key point of removal is to eliminate the interference information highly correlated with the moisture factor. The removal process can be to first extract the main features of the spectral data, and then perform orthogonal calculation on the features and the reference matrix to eliminate the feature information related to the reference matrix in the features, obtaining corrected spectral data. In this way, the inherent spectral features of the target can be more accurately reflected, providing high-quality data for subsequent hyperspectral information extraction and quantitative remote sensing applications.

[0057] Finally, calculate the correlation between the corrected spectral data and the standard spectral library. The standard spectral library stores a large amount of normal spectral data that is not affected by external factors and has no moisture content. By calculating the correlation coefficient between each group of corrected spectra and the spectra in the standard spectral library, the spectrum with the strongest correlation can be selected as the final corrected spectrum after removing the influence of moisture.

[0058] In the embodiments of the present invention, directly using the preprocessed original hyperspectral reflectance data, calculate and analyze the moisture factor that can accurately characterize the moisture content of the target object based on the short-wave infrared band, without relying on external measured auxiliary data, greatly simplifying the data acquisition and processing process. Only by using the moisture factor calculated from the hyperspectral reflectance data as a reference can the influence of moisture be removed, significantly reducing the dependence on external measured data, improving the applicability of the subsequent spectral application model, and at the same time being able to keep the inherent characteristics of the ground object spectrum unchanged, and can provide technical support for the accurate application of hyperspectral data in multi-temporal, multi-scene, and large-scale quantitative remote sensing, information extraction, target recognition and other fields.

[0059] In some embodiments, spectral denoising preprocessing is performed on the original hyperspectral reflectance data according to the noise level to obtain the target hyperspectral reflectance data, which can be achieved through the following process.

[0060] First, it is necessary to remove the high-noise band spectral data of the original hyperspectral reflectance data affected by water vapor to obtain the spectral data to be processed. For example, the high-noise bands of the ground object in the range of 1400nm to 1900nm or nearby. The original hyperspectral reflectance data of these bands is severely affected by water vapor and cannot accurately retain spectral information, so it can be directly removed.

[0061] Next, divide the spectral data to be processed into spectral data of high-noise bands, low-noise bands, and medium-noise bands according to the signal-to-noise ratio. Here, different noise levels can be divided according to the signal-to-noise ratio, and according to this noise level, the spectral data to be processed can be divided into spectral data of high-noise bands, low-noise bands, and medium-noise bands.

[0062] Finally, perform denoising processing on the spectral data of high-noise bands, low-noise bands, and medium-noise bands respectively to obtain the target hyperspectral reflectance data. Among them, the denoising process can be: for the spectral data to be processed in the high-noise band, use the wavelet threshold denoising method for denoising, for the spectral data to be processed in the low-noise band, use the moving average algorithm for denoising, and for the spectral data to be processed in the medium-noise band, use the Savitzky-Golay filtering algorithm for denoising. Finally, combine the spectral data after denoising of each part of the band to obtain the target hyperspectral reflectance data.

[0063] In the embodiments of the present invention, by performing spectral denoising preprocessing on the original hyperspectral reflectance data and further using different denoising algorithms for denoising according to different noise bands, more spectral information can be retained to the greatest extent, the influence of external noise factors on the spectral data can be effectively removed, and the quality of the spectral data can be improved.

[0064] In some embodiments, a moisture factor is calculated based on the target hyperspectral reflectance data. Specifically, first, a short-wave infrared band that is sensitive to moisture content is determined. The range of this short-wave infrared band is generally 1000nm - 2500nm. Within this band range, the moisture content has higher sensitivity, so this short-wave infrared band is selected to calculate the moisture factor.

[0065] Then, for the target hyperspectral reflectance data, according to the short-wave infrared band, the short-wave infrared moisture stress index, the normalized short-wave infrared moisture index, and the visible and short-wave infrared dry index are calculated respectively as the moisture factor. In actual implementation, the calculated moisture factors include but are not limited to the short-wave infrared moisture stress index, the normalized short-wave infrared moisture index, and the visible and short-wave infrared dry index, which should depend on the specific scenario.

[0066] For the short-wave infrared moisture stress index (SIWSI), the calculation formula is as follows:

[0067] (1)

[0068] In the above formula (1), represents the reflectance of the short-wave infrared band (such as 1000nm - 2500nm) in the target hyperspectral reflectance data, represents the reflectance of the near-infrared band (such as 700nm - 1000nm) in the target hyperspectral reflectance data.

[0069] For the normalized short-wave infrared moisture index (NSWI), the calculation formula is as follows:

[0070] (2)

[0071] In the above formula (2), represents the reflectance of the band with a wavelength of 1550nm - 1750nmnm in the target hyperspectral reflectance data, represents the reflectance of the band with a wavelength of 2080nm - 2350nm in the target hyperspectral reflectance data.

[0072] For the visible and short-wave infrared dry index (VSDI), the calculation formula is as follows:

[0073] (3)

[0074] In the above formula (3), represents the reflectance of the short-wave infrared band (such as 1000nm - 2500nm) in the target hyperspectral reflectance data, represents the reflectance of the red light band (such as 627nm - 700nm) in the target hyperspectral reflectance data, represents the reflectance of the blue light band (such as 446nm - 524nm) in the target hyperspectral reflectance data.

[0075] When calculating the above moisture factor, all band combinations within the involved wavelength range are traversed and calculated one by one. In addition, for different target objects and application scenarios, the wavelength range used to calculate the moisture factor can be dynamically adjusted, effectively improving the practicability and stability of the moisture factor in different scenarios.

[0076] After calculating the moisture factor, a corresponding reference matrix Y needs to be constructed to characterize the degree of moisture influence. When constructing the reference matrix Y, one of the short-wave infrared moisture stress index, normalized short-wave infrared moisture index, and visible and short-wave infrared drying index can be selected as the moisture factor to directly construct the corresponding reference matrix Y. It is also possible to construct the corresponding matrices for the three indices respectively, and then set three corresponding matrix weights to weight the matrices. The three matrix weights can be the same or different, depending on the specific situation. Finally, the three weighted matrices are combined to obtain the final reference matrix Y. The construction method can be a linear mapping to visualize the index as a matrix.

[0077] In the embodiments of the present invention, the short-wave infrared moisture stress index, normalized short-wave infrared moisture index, and visible and short-wave infrared drying index are calculated simultaneously as three moisture factors, and all band combinations within the involved wavelength range are traversed and calculated one by one. In addition, for different target objects and application scenarios, the wavelength range used to calculate the moisture factor can be dynamically adjusted, which can more accurately characterize the moisture content in the target hyperspectral reflectance data and facilitate subsequent moisture removal.

[0078] In some embodiments, after calculating the moisture factor as a reference matrix, it is necessary to remove the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain the corrected spectral data. This process can be achieved through the following steps 201 to 203, which are specifically described below.

[0079] Step 201: Use the singular value decomposition algorithm to extract the eigencomponents from the spectral data matrix corresponding to the target hyperspectral reflectance data.

[0080] Here, the Singular Value Decomposition (SVD) algorithm is used to extract feature components from the spectral data matrix X corresponding to the target hyperspectral reflectance data. These feature components can characterize the main features of the target hyperspectral reflectance data. Of course, these main features are affected by moisture. And the information features related to the reference matrix Y are removed from it, and the calculation is based on these main features.

[0081] Step 202: Construct a projection matrix in the orthogonal direction of the spectral data matrix with respect to the reference matrix.

[0082] Based on the spectral data matrix X corresponding to the target hyperspectral reflectance data, a projection matrix in the orthogonal direction of the reference matrix Y is constructed, denoted as PROJ, and is expressed by the following formula:

[0083] (4)

[0084] In the above formula (4), I is the identity matrix and Y is the reference matrix.

[0085] Step 203: Perform feature mapping of the feature components in the projection matrix to obtain corrected spectral data.

[0086] Finally, the feature components extracted in step 201 are subjected to feature mapping in the projection matrix PROJ, and the corresponding mapping features are calculated therefrom. The process of feature mapping can be to directly perform matrix multiplication on the matrix corresponding to the feature components and the projection matrix. The obtained mapping feature is the result after removing the information features related to the reference matrix Y from the feature components. Based on this result, the corresponding corrected spectral data can be restored.

[0087] In the embodiments of the present invention, only by using the moisture factor calculated from the hyperspectral reflectance data as a reference, the removal of the influence of moisture in the spectral data can be achieved, significantly reducing the dependence on external measured data, simplifying the spectral processing process, and improving the applicability of subsequent spectral application models. At the same time, it can keep the inherent characteristics of the ground object spectrum unchanged, and can provide technical support for the accurate application of hyperspectral data in multiple fields such as multi-temporal, multi-scene, large-scale quantitative remote sensing, information extraction, and target recognition.

[0088] In some embodiments, the removal of the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data is achieved through an adaptive iterative optimization strategy. Because in the actual environment, only by the removal steps of steps 201 to 203 above to obtain the corrected spectral data, the moisture features therein often cannot be completely and effectively removed. Therefore, in the embodiments of the present invention, an adaptive iterative optimization strategy is introduced here to remove the influence of moisture in the target hyperspectral reflectance data.

[0089] In the adaptive iterative optimization strategy, it is first necessary to preset a convergence threshold for removing the influence of moisture. This convergence threshold represents the effective standard that the target hyperspectral reflectance data should reach when removing the influence of moisture. When this convergence threshold is reached, it indicates that the moisture characteristics of the target hyperspectral reflectance have been effectively removed. The convergence threshold can be determined from the normal spectral data that is not affected by moisture.

[0090] According to the adaptive iterative optimization strategy, after calculating the corrected spectral data using the above steps 201 to 203, it is necessary to determine whether the current corrected spectral data meets the convergence threshold for removing the influence of moisture. If the current corrected spectral data does not reach the convergence threshold, the following process is iteratively executed: Remove the information features related to the reference matrix from the current corrected spectral data. That is, based on the current corrected spectral data after moisture, continue to execute the moisture removal process of the above steps 201 to 203. In each iteration process, the correlation between the spectral data matrix corresponding to the current corrected spectral data and the reference matrix Y can be gradually weakened. Through this continuous iteration process, the moisture interference information is gradually reduced until, after a certain iteration process is completed, the current corrected spectral data obtained reaches the preset convergence threshold.

[0091] If the current corrected spectral data reaches the convergence threshold, the iteration ends. By iteratively executing the moisture removal steps of the above steps 201 to 203, the obtained corrected spectral data will gradually approach and meet the convergence threshold. When the convergence threshold is reached, it indicates that the effective removal of moisture has been completed at this time, and the iteration can be stopped.

[0092] In the embodiment of the present invention, the adaptive iterative optimization strategy is adopted, and a convergence threshold for removing the influence of moisture is preset to ensure that the target hyperspectral reflectance data can reach the standard of moisture removal, further improving the effectiveness of removing the influence of moisture from the corrected spectral data.

[0093] In some embodiments, after completing the removal of the influence of moisture, multiple corrected spectral features will be obtained. Therefore, it is necessary to calculate the correlation between the corrected spectral data and the standard spectral library to obtain the final corrected spectral data after removing the influence of moisture. In the embodiment of the present invention, the corrected spectral data is further selected through the standard spectral library. The standard spectral library stores standard spectral data that is not affected by moisture. Using these spectral data as templates, the corrected spectral data can be further optimized, and the spectral data with the strongest correlation is selected as the final corrected spectral data after removing the influence of moisture.

[0094] The specific process is as follows: First, calculate the correlation coefficient between the calibrated spectral data and the spectra in the standard spectral library. The larger the correlation coefficient between the spectra, the closer the calibrated spectral data is to the template. Calculating the correlation coefficient between the spectra can be the similarity of spectral features. For example, first extract the spectral features of the calibrated spectral data and the template respectively, and then calculate the similarity of the spectral features. The similarity can be the Euclidean distance or the cosine distance, etc., which is not limited here. The larger the similarity, the larger the correlation coefficient between the spectra.

[0095] Calculate the correlation coefficient between the spectra for each of the obtained calibrated spectral features and the standard spectral library one by one. Finally, select the calibrated spectral data with the largest correlation coefficient between the spectra as the final calibrated spectrum after removing the influence of moisture. That is to say, select the calibrated spectral data with the highest similarity of spectral features as the final calibrated spectrum after removing the influence of moisture.

[0096] In the embodiment of the present invention, after moisture removal, the calibrated spectral data is further optimized through the standard spectral library, which can further ensure the accurate elimination of the influence of moisture on the spectral data, effectively improve the quality of the spectral data, and provide technical support for the precise application of hyperspectral data in multiple fields such as multi-temporal, multi-scenario, large-scale quantitative remote sensing, information extraction, and target recognition.

[0097] Next, the device for removing the influence of moisture on hyperspectral reflectance data provided by the present invention will be described. The device for removing the influence of moisture on hyperspectral reflectance data described below can be mutually referred to corresponding to the method for removing the influence of moisture on hyperspectral reflectance data described above.

[0098] As Figure 2 shown, the device for removing the influence of moisture on hyperspectral reflectance data provided by the present invention specifically includes: an acquisition module 201, a preprocessing module 202, a calculation module 203, and a removal module 204. Among them, the acquisition module 201 is used to acquire the original hyperspectral reflectance data of the ground object in its natural state; the preprocessing module 202 is used to perform spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain the target hyperspectral reflectance data; the calculation module 203 is used to calculate the moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data; the removal module 204 is used to use the matrix composed of the moisture factors as the reference matrix, remove the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain the calibrated spectral data, and perform a correlation calculation on the calibrated spectral data and the standard spectral library to obtain the final calibrated spectrum after removing the influence of moisture.

[0099] It should be noted that the beneficial effects of the moisture influence removal device for hyperspectral reflectance data correspond to those of the moisture influence removal method for hyperspectral reflectance data in the above text. Therefore, the beneficial effects of the moisture influence removal device for hyperspectral reflectance data will not be elaborated here.

[0100] Figure 3 An example of the physical structure diagram of an electronic device is shown as Figure 3 shown. The electronic device may include: a processor 310, a communication interface 320, a memory 330, and a communication bus 340. Among them, the processor 310, the communication interface 320, and the memory 330 complete communication with each other through the communication bus 340. The processor 310 can call the logical instructions in the memory 330 to execute the method for removing the moisture influence of hyperspectral reflectance data, and the method includes: obtaining the original hyperspectral reflectance data in the natural state of the ground object; performing spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain the target hyperspectral reflectance data; calculating the moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data; using the matrix composed of the moisture factors as the reference matrix, removing the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain the corrected spectral data, and performing a correlation calculation on the corrected spectral data and the standard spectral library to obtain the final corrected spectral data after removing the moisture influence.

[0101] In addition, when the logical instructions in the above-mentioned memory 330 are implemented in the form of software functional units and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0102] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the method for removing the moisture influence on hyperspectral reflectance data provided by each of the above methods. The method includes: obtaining the original hyperspectral reflectance data of a ground object in its natural state; performing spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain target hyperspectral reflectance data; calculating a moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data; using the matrix composed of the moisture factors as a reference matrix, removing the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain corrected spectral data, and performing a correlation calculation on the corrected spectral data and a standard spectral library to obtain the final corrected spectrum after removing the moisture influence.

[0103] In another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it realizes the method for removing the moisture influence on hyperspectral reflectance data provided by each of the above methods. The method includes: obtaining the original hyperspectral reflectance data of a ground object in its natural state; performing spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain target hyperspectral reflectance data; calculating a moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data; using the matrix composed of the moisture factors as a reference matrix, removing the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain corrected spectral data, and performing a correlation calculation on the corrected spectral data and a standard spectral library to obtain the final corrected spectrum after removing the moisture influence.

[0104] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative efforts.

[0105] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for removing the influence of moisture on hyperspectral reflectance data, characterized in that The method includes: Obtaining the original hyperspectral reflectance data of the ground object in its natural state; Performing spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain target hyperspectral reflectance data; Calculating a moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data; Using the matrix composed of the moisture factors as a reference matrix, removing the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain corrected spectral data, and performing a correlation calculation between the corrected spectral data and a standard spectral library to obtain the final corrected spectrum after removing the influence of moisture; The removing the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain corrected spectral data includes: Extracting eigencomponents from the spectral data matrix corresponding to the target hyperspectral reflectance data by using the singular value decomposition algorithm; Constructing a projection matrix of the spectral data matrix in the orthogonal direction of the reference matrix; Performing eigenmapping of the eigencomponents in the projection matrix to obtain corrected spectral data; The calculating a moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data includes: Determining the short-wave infrared bands sensitive to moisture content; For the target hyperspectral reflectance data, calculating the short-wave infrared moisture stress index, the normalized short-wave infrared moisture index, and the visible and short-wave infrared dry index respectively according to the short-wave infrared bands as moisture factors.

2. The method for removing the influence of moisture on hyperspectral reflectance data according to claim 1, wherein The performing spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain target hyperspectral reflectance data includes: Removing the spectral data of the high-noise bands affected by water vapor in the original hyperspectral reflectance data to obtain the spectral data to be processed; Dividing the spectral data to be processed into spectral data of high-noise bands, low-noise bands, and medium-noise bands according to the signal-to-noise ratio; Performing denoising processing on the spectral data of high-noise bands, low-noise bands, and medium-noise bands respectively to obtain target hyperspectral reflectance data; Among them, the process of denoising processing includes: For the spectral data to be processed in high-noise bands, using the wavelet threshold denoising method for denoising; For the spectral data to be processed in low-noise bands, using the moving average algorithm for denoising; For the spectral data to be processed in medium-noise bands, using the Savitzky-Golay filtering algorithm for denoising.

3. The method for removing the influence of moisture on hyperspectral reflectance data according to claim 1, wherein The removing the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data is realized by an adaptive iterative optimization strategy, and the adaptive iterative optimization strategy includes: Presetting a convergence threshold for removing the influence of moisture; If the current corrected spectral data does not reach the convergence threshold, then iteratively execute the following process: Removing the information features related to the reference matrix from the current corrected spectral data; If the current corrected spectral data reaches the convergence threshold, then end the iteration.

4. The method for removing the influence of moisture on hyperspectral reflectivity data according to claim 1, characterized in that The performing a correlation calculation between the corrected spectral data and a standard spectral library to obtain the final corrected spectrum after removing the influence of moisture includes: Calculate the correlation coefficient between the corrected spectral data and the spectra in the standard spectral library; Select the corrected spectral data with the largest correlation coefficient between spectra as the final corrected spectrum after removing the influence of moisture.

5. A device for removing the influence of moisture on hyperspectral reflectance data, characterized in that, The device includes: An acquisition module for acquiring the original hyperspectral reflectance data of the ground object in its natural state; A preprocessing module for performing spectral denoising preprocessing on the original hyperspectral reflectance data according to the noise level to obtain the target hyperspectral reflectance data; A calculation module for calculating the moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data; A removal module for removing the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data with the matrix composed of the moisture factors as the reference matrix to obtain the corrected spectral data, and performing a correlation calculation on the corrected spectral data and the standard spectral library to obtain the final corrected spectrum after removing the influence of moisture; The removing the information features related to the reference matrix from the spectral data matrix corresponding to the target hyperspectral reflectance data to obtain the corrected spectral data includes: Extracting the eigencomponents from the spectral data matrix corresponding to the target hyperspectral reflectance data by using the singular value decomposition algorithm; Constructing the projection matrix of the spectral data matrix in the orthogonal direction of the reference matrix; Performing eigenmapping on the eigencomponents in the projection matrix to obtain the corrected spectral data; The calculating the moisture factor related to the short-wave infrared band based on the target hyperspectral reflectance data includes: Determining the short-wave infrared bands that are sensitive to the moisture content; For the target hyperspectral reflectance data, calculating the short-wave infrared moisture stress index, the normalized short-wave infrared moisture index, and the visible and short-wave infrared dryness index respectively according to the short-wave infrared bands as the moisture factors.

6. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and running on the processor, wherein, When the processor executes the computer program, it implements the method for removing the influence of moisture on the hyperspectral reflectance data according to any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for removing the influence of moisture on the hyperspectral reflectance data according to any one of claims 1 to 4.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for removing the influence of moisture on the hyperspectral reflectance data according to any one of claims 1 to 4.

Images