Building house remote sensing surveying and mapping data processing method and system

Abstract

The invention relates to the technical field of remote sensing surveying and mapping, in particular to a building house remote sensing surveying and mapping data processing method and system. The method comprises the following steps: acquiring hyperspectral data of a to-be-measured region at different sampling moments, and combining fluctuation conditions of wavelength absorptivity at different sampling points with vegetation density difference conditions to obtain density difference parameters; the contribution degree of spectral data at different sampling points to the density of the building house is quantified, the difference of wavelength absorptivity is analyzed on the basis of density difference parameters, and the retention degree value of each wavelength is obtained and used for evaluating the importance of each wavelength. The initial covariance matrix is corrected by using the information, the weight of the key wavelength is improved, and an updated covariance matrix is obtained. Finally, dimension reduction analysis is carried out on the updated covariance matrix based on principal component analysis, it is ensured that a dimension reduction result can more accurately reflect the building house density condition, and more valuable information is provided for city management.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing mapping, and particularly to a method and system for processing remote sensing mapping data of building houses. Background Art

[0002] With the rapid development of remote sensing technology, hyperspectral imaging technology has become an important means for obtaining information about building houses in the field of mapping. Hyperspectral data provides rich spectral information due to its high-dimensional characteristics. However, the dimension of hyperspectral data is usually very high. For example, there may be hundreds or even thousands of wavelength data for each location point. Therefore, the dimension of hyperspectral data can be reduced to extract the core information of the original data, thereby improving the efficiency and accuracy of remote sensing monitoring of building houses.

[0003] When the prior art reduces the dimension of hyperspectral data, it usually performs undifferentiated dimension reduction on the hyperspectral data using the principal component analysis method. However, due to various environmental factors at different times, the absorption rate of wavelengths in the hyperspectral data will be affected. If undifferentiated dimension reduction is used, too much key information will be lost in the dimension reduction result, resulting in the dimension reduction result being unable to accurately represent the density status of the building house itself. Summary of the Invention

[0004] In order to solve the technical problem that undifferentiated dimension reduction will cause too much key information to be lost in the dimension reduction result of hyperspectral data of building houses, resulting in the dimension reduction result being unable to accurately represent the density status of the building house itself, the purpose of the present invention is to provide a method and system for processing remote sensing mapping data of building houses. The specific technical solutions adopted are as follows: Obtain hyperspectral data of the area to be measured at different sampling times; For the hyperspectral data at any sampling time, according to the differences in the vegetation density between different sampling points, the differences in the absorption rates of wavelengths between different sampling points, and the fluctuation of the absorption rates of all sampling points at the same wavelength, obtain the density difference parameter between different sampling points in the hyperspectral data at this sampling time; For any sampling point in the area to be measured, fuse and analyze the positional relationship, the differences in spectral data, and the density difference parameter between this sampling point and the remaining other sampling points at all sampling times to obtain the abnormal degree value of the building house density at this sampling point; for any wavelength, fuse and analyze the differences in the absorption rates of this wavelength at different times of all sampling points and the abnormal degree value of the building house density of each sampling point, and then analyze the fluctuation situation to obtain the retention degree value of this wavelength; Based on the hyperspectral data of all sampling points at all sampling times, the initial covariance matrix corresponding to the spectral data is calculated; the element values in the initial covariance matrix are corrected according to the retention degree values of each wavelength to obtain an updated covariance matrix; principal component analysis is performed on the updated covariance matrix to obtain a dimensionality reduction result.

[0005] Further, the method for obtaining the density difference parameter includes: Obtain the normalized difference vegetation index at each sampling point in the area to be measured. At any sampling time, the difference in the normalized difference vegetation index between two different sampling points is used as the vegetation condition difference value. Analyze the difference between the absorption rates of two different sampling points at the same wavelength and the fluctuation of the absorption rates of all sampling points at the same wavelength to obtain the density difference factor between different sampling points at this sampling time. According to the vegetation condition difference value and the density difference factor between two different sampling points, the density difference parameter between the two different sampling points at this sampling time is obtained. The density difference parameter is positively correlated with the density difference factor, negatively correlated with the vegetation condition difference value, and the value of the density difference parameter is a normalized value.

[0006] Further, the method for obtaining the density difference factor includes: Optionally select a wavelength as the wavelength to be measured. At this sampling time, the difference in the absorption rate of two different sampling points at the wavelength to be measured is used as the absorption rate difference value; the standard deviation of the absorption rates of all sampling points at the wavelength to be measured is used as the absorption rate fluctuation value. According to the absorption rate fluctuation value and the absorption rate difference value, obtain the absorption rate change index of two sampling points at the wavelength to be measured. The absorption rate fluctuation value and the absorption rate difference value both change in the same direction as the absorption rate change index; the average value of the absorption rate change indexes of two different sampling points at all wavelengths is used as the density difference factor.

[0007] Further, the method for obtaining the abnormal degree value of the building housing density includes: Optionally select a sampling point as the target point. According to the distance between the target point and each remaining sampling point, screen out the comparison points of the target point among the remaining sampling points. For any sampling moment, the difference in the absorbance of the same wavelength in the spectral data between the target point and each comparison point is used as the absorbance distance value; the sum value of the absorbance distance values of all wavelengths is used as the sum distance value. According to the density difference parameter between the target point and each comparison point and the sum distance value, the deviation value of the spectral data between the target point and each comparison point at this sampling moment is obtained. The deviation value is negatively correlated with the density difference parameter, and the sum distance value and the deviation value are positively correlated; The mean value of the deviation values of the spectral data between the target point and each comparison point at all sampling moments is used as the abnormal factor of the building housing density at the target point; the mean value of the abnormal factors of the building housing density between the target point and all comparison points is used as the abnormal degree value of the building housing density at the target point.

[0008] Further, the method for obtaining the comparison points includes: Calculate the Euclidean distance between the target point and the position of each remaining sampling point as the distance factor, and use the sampling points with the distance factor less than the preset distance threshold as the comparison points of the target point.

[0009] Further, the method for obtaining the retention degree value includes: Combine all sampling moments in pairs to obtain all non-repeating moment combinations; For any sampling point, in each moment combination, fuse the abnormal degree value of the building housing density at this sampling point with the absorbance difference of the wavelength to be measured at this sampling point at two sampling moments to obtain the absorbance change characteristic value of this sampling point; The standard deviation of the absorbance change characteristic values of the wavelength to be measured at the same moment combination for all sampling points is used as the retention factor of the wavelength to be measured at the same moment combination; the mean value of the retention factors of the wavelength to be measured at all moment combinations is used as the retention degree value of the wavelength to be measured.

[0010] Further, the method for obtaining the absorbance change characteristic value includes: The value obtained by performing negative correlation mapping and normalization on the abnormal degree value of the building housing density at each sampling point is used as the retention weight; for any sampling point, in each moment combination, the absorbance difference of the wavelength to be measured at this sampling point at two sampling moments is used as the absorbance change factor; the product of the absorbance change factor of this sampling point at each moment combination and the corresponding retention weight is used as the absorbance change characteristic value of this sampling point at the wavelength to be measured.

[0011] Further, the method for obtaining the initial covariance matrix includes: The hyperspectral data includes two position data and one spectral data; Take each spectral data as a column vector, where the length of the column vector is equal to the number of wavelengths, and the element value is the wavelength absorption rate; Form an initial matrix with the column vectors corresponding to the spectral data of all sampling points at all sampling times, and calculate the initial matrix to obtain the initial covariance matrix.

[0012] Furthermore, the method for obtaining the updated covariance matrix includes: Construct a correction matrix based on the retention degree values of each wavelength. The correction matrix is a diagonal matrix, and each element value on the diagonal is the retention degree value of each wavelength, and the element values in other positions are 0; Multiply the matrix obtained by multiplying the correction matrix by the initial covariance matrix by the correction matrix again to obtain the updated covariance matrix.

[0013] The present invention also proposes a remote sensing mapping data processing system for building houses, and the system includes: A memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of any one of the methods can be implemented.

[0014] The present invention has the following beneficial effects: The object of the present invention is to improve the importance degree of the wavelength reflecting the building houses, so as to improve the accuracy of the dimensionality reduction result representing the density condition of the building houses in the subsequent dimensionality reduction process. First, hyperspectral data of the area to be measured at different sampling times is obtained. Since even at the same sampling time, if only the difference in wavelength absorption rates between sampling points is analyzed, it is impossible to well distinguish whether the difference is affected by the density of vegetation. Therefore, when analyzing the difference in wavelength absorption rates between different sampling points and the fluctuation of absorption rates of all sampling points at the same wavelength, the difference in the density of vegetation between different sampling points is combined therewith to obtain the building house density difference parameter between different sampling points at each sampling time. This parameter more accurately reflects the difference in building house density between different sampling points under the influence of the density of vegetation. Further, in order to determine whether each wavelength determines the density of the building houses, it is necessary to analyze the contribution degree of the spectral data of different sampling points to the density of the building houses. Since the difference in spectral data of sampling points with close positions may be caused by the different density of roads between the building houses, on the basis of the above-mentioned building house density difference parameter, the position relationship between each sampling point and the difference between the spectral data are combined to obtain the building house density anomaly degree value at each sampling point, which is used to quantify the contribution degree of the spectral data at this sampling point to the density of the building houses. Further, by performing a fusion analysis on the difference in absorption rates of each wavelength of different sampling points at different times and the building house density anomaly degree value and evaluating its fluctuation, the retention degree value of each wavelength can be obtained, which is used to distinguish the wavelengths with high research value and low research value for the density condition of the building houses. Then, based on all the hyperspectral data, the initial covariance matrix is calculated, and the element values in the initial covariance matrix are corrected based on the retention degree values of each wavelength, so that the wavelengths with high retention degree values have higher weights, and the updated covariance matrix is obtained. Finally, the updated covariance matrix is continuously subjected to principal component analysis, and the obtained dimensionality reduction result will contain more key information and can more accurately represent the density condition of the building houses. Description of the Drawings

[0015] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to these drawings.

[0016] Figure 1 It is a flowchart of a method for processing remote sensing mapping data of building houses provided by an embodiment of the present invention; Figure 2Flowchart of a method for obtaining density difference parameters provided by an embodiment of the present invention; Figure 3 Flowchart of a method for obtaining the abnormal degree value of building density provided by an embodiment of the present invention; Figure 4 System block diagram of a building remote sensing mapping data processing system provided by an embodiment of the present invention; Figure 5 Schematic diagram of a system provided by an embodiment of the present invention; Reference numerals: 401, data acquisition module; 402, density difference analysis module; 403, wavelength retention degree analysis module; 404, dimensionality reduction module; 500, processor; 501, memory; 502, bus; 503 communication interface. Detailed implementation manners

[0017] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following combines the accompanying drawings and preferred embodiments to detail the specific implementation manners, structures, features and effects of a building remote sensing mapping data processing method and system proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0019] The following specifically describes the specific solutions of a building remote sensing mapping data processing method and system provided by the present invention with reference to the accompanying drawings.

[0020] Please refer to Figure 1 , which shows the flowchart of a building remote sensing mapping data processing method provided by an embodiment of the present invention. The method includes the following steps: Step S1: Obtain hyperspectral data of the area to be measured at different sampling times.

[0021] With the rapid development of remote sensing technology, hyperspectral imaging technology has evolved into a core tool for obtaining building information in the field of surveying and mapping. Although hyperspectral data provides rich spectral information due to its high-dimensional characteristics, it also brings challenges such as large data volume, high information redundancy, and increased computational complexity. Machine learning technology has achieved remarkable results in the field of data mining, especially showing strong potential in feature extraction and dimensionality reduction. Therefore, the dimensionality reduction method based on machine learning can automatically extract the most representative feature information by learning the internal laws and features in the original data, thereby retaining the core information of the original data while reducing the data dimension. This method can be applied to the dimensionality reduction process of building hyperspectral data.

[0022] First, aerial photography can be used to obtain hyperspectral data of the area to be measured at different sampling times using a hyperspectral imager. The interval between sampling times is set to 2 hours, the wavelength range of the spectrum is 400 - 1000 nanometers, and the wavelength resolution is set to 1 nanometer.

[0023] It should be noted that hyperspectral data contains three-dimensional data. Two of the spatial dimensions can be used to determine the position of the sampling points, and the spectral dimension contains the spectral data of the sampling points. The horizontal axis of the spectral data is the wavelength, and the vertical axis is the absorption rate. The specific acquisition method, wavelength range, resolution, interval between sampling times, etc. of the hyperspectral data can all be adjusted according to the implementation scenario and are not limited here.

[0024] Step S2: For the hyperspectral data at any sampling time, based on the differences in the vegetation density between different sampling points, the differences in the absorption rates of wavelengths at different sampling points, and the fluctuations in the absorption rates of all sampling points at the same wavelength, obtain the difference parameter of the building density between different sampling points in the hyperspectral data at this sampling time.

[0025] In the embodiment of the present invention, the principal component analysis method is used for data dimensionality reduction. However, since not every wavelength in the spectral data can well reflect the building density, if indiscriminate dimensionality reduction is performed on all wavelengths in the spectral data, the final dimensionality reduction result will not be able to well represent the building density. Therefore, different wavelengths need to be distinguished.

[0026] Since the hyperspectral data of the same area to be measured is collected at different times, the sampling points in the hyperspectral data at different times will correspond one by one. That is, the same sampling point will have spectral data at multiple different times. It should be noted that the SIFT feature point matching method can also be used to match the sampling points in the hyperspectral data at different times to obtain the spectral data at the same sampling point at different times. And the SIFT feature point matching algorithm is a well-known technical means in the art, so it will not be elaborated here.

[0027] Because in the hyperspectral data of the area to be measured, there will be some vegetation parts without buildings at some sampling points, so just based on the difference in spectral data between two points, it is impossible to better distinguish whether the difference in spectral data is caused by the different densities of buildings or the density of vegetation. Therefore, considering the difference in the density of vegetation between sampling points, then analyzing the difference in wavelength absorption rate between sampling points, and the fluctuation of the absorption rate of all sampling points at the same wavelength, so as to obtain the density difference parameter between different sampling points, which is used to measure the difference in the density of building houses between sampling points.

[0028] Preferably, in an embodiment of the present invention, the method for obtaining the density difference parameter includes: Please refer to Figure 2 , which shows the flowchart of a method for obtaining a density difference parameter provided by an embodiment of the present invention. The method includes the following steps: Step S201: Based on the difference in the density of vegetation at different sampling points in the area to be measured, determine the difference value of the vegetation status between different sampling points.

[0029] According to the spectral characteristics of vegetation, the satellite visible light and near-infrared bands are combined to form various vegetation indices. Vegetation indices are simple, effective, and empirical measures of the surface vegetation status. Therefore, the normalized vegetation index at each sampling point in the area to be measured can be obtained , which is used to quantify the density of vegetation at each sampling point.

[0030] Then, at any sampling time, the difference between the normalized vegetation indices between two different sampling points is used as the difference value of the vegetation status. The formula model of the difference value of the vegetation status is: Where represents at the th sampling time, the difference value of the vegetation status between sampling point and sampling point ; represents the th sampling time, sampling point Normalized Difference Vegetation Index (NDVI) at Indicates the th sampling moment, the normalized difference vegetation index at the sampling point

[0031] In the formula model of the vegetation condition difference value, if there is a significant difference in the vegetation density at two sampling points, for example, the vegetation is dense at one sampling point while the other sampling point belongs to the construction area, then the vegetation condition difference value between these two sampling points will be larger; conversely, if the vegetation density at the two sampling points is relatively consistent, then the vegetation condition difference value will be correspondingly smaller.

[0032] It should be noted that the method for obtaining the normalized difference vegetation index is a well-known technical means to those skilled in the art, and will not be elaborated here. Moreover, in other embodiments of the present invention, when calculating the vegetation condition difference value, the normalized difference vegetation index can also be replaced by Ratio Vegetation Index (RVI), Green Vegetation Index (GVI), etc., and the obtaining methods are all processes well-known to those skilled in the art, and will not be limited and elaborated here.

[0033] Step S202: Analyze the difference between the absorptions of different sampling points at the same wavelength and the fluctuation of the absorptions of all sampling points at the same wavelength to obtain the density difference factor between different sampling points at this sampling moment.

[0034] To quantify the density difference between different sampling points, an arbitrary wavelength can be first selected as the wavelength to be measured.

[0035] Then, at this sampling moment, the difference between the absorptions of two different sampling points at the wavelength to be measured is used as the absorption difference value; the standard deviation of the absorptions of all sampling points at the wavelength to be measured is used as the absorption fluctuation value. Both the absorption difference value and the absorption fluctuation value characterize the change of the absorptions of the sampling points at the same wavelength. Therefore, by combining the two, an absorption change index of the two sampling points at the wavelength to be measured is obtained based on the absorption fluctuation value and the absorption difference value, and both the absorption fluctuation value and the absorption difference value vary in the same direction as the absorption change index.

[0036] Finally, the average value of the absorption change indexes of two different sampling points at all wavelengths is used as the density difference factor. The formula model of the density difference factor can be specifically, for example: Wherein, Indicates the density difference factor between the sampling point and the sampling point at the th sampling moment; Indicates the total number of wavelengths; ​Represents the absorption rate of the wavelength; Represents at the th wavelength, the absorption rate fluctuation value of all sampling points; Represents at the th sampling moment, the absorption rate of the sampling point at the th wavelength; Represents at the th sampling moment, the absorption rate of the sampling point at the th wavelength.

[0037] In the formula model of the density difference factor, at a certain sampling moment, analyzing globally, calculate the standard deviation of the absorption rates of all sampling points at the same wavelength , as the absorption rate fluctuation value, used to measure the change of the absorption rate at the same wavelength. The larger this value is, it indicates that overall, there are relatively significant differences in the absorption rates of all sampling points at this wavelength. Then, there is a greater possibility of differences in the density corresponding to this wavelength between two sampling points. Then, analyzing from a single wavelength locally, calculate the difference in the absorption rates of two sampling points at the same wavelength, as the absorption rate difference value . The larger this value is, it indicates that there are significant differences in the density corresponding to this wavelength between the two sampling points. Then, combine the absorption rate fluctuation value at each wavelength and the absorption rate difference value between the two sampling points, and take the product of the two as the absorption rate change index of the two sampling points at each wavelength . The larger this index is, it indicates that there are significant density differences between the two sampling points for the substance corresponding to this wavelength. Finally, take the average value of the absorption rate change indexes of the two sampling points at all wavelengths as the density difference factor between the two sampling points. The larger the density difference factor is, it indicates that there are significant differences in the density between the two sampling points.

[0038] Since both the absorption rate fluctuation value and the absorption rate difference value change in the same direction as the absorption rate change index, in other embodiments of the present invention, the absorption rate change index can also be expressed as , and the specific calculation method is not limited here.

[0039] Step S203: Obtain the density difference parameter between different sampling points at this sampling moment according to the vegetation condition difference value and the density difference factor between different sampling points.

[0040] Since the density difference factor only analyzes the differences in the absorption rates of different wavelengths in the spectral data of different sampling points, and there are obvious differences in the spectral data of sampling points with different vegetation densities, in order to reduce the impact of this situation, the vegetation condition difference value obtained in step S201 and the density difference factor obtained in step S202 should be combined to obtain the density difference parameter between different sampling points. The formula model of the density difference parameter can be specifically, for example: Wherein, represents the density difference parameter between sampling point and sampling point at the th sampling moment; represents the vegetation condition difference value between sampling point and sampling point at the th sampling moment; represents the density difference factor between sampling point and sampling point at the th sampling moment; represents a preset first parameter; represents a normalization function.

[0041] In the formula model of the density difference parameter, when there are significant differences in the vegetation density at two sampling points, the vegetation condition difference value at these two sampling points will be larger, and then the difference in density between these two sampling points is more likely to be caused by the difference in vegetation coverage. That is, if the vegetation condition difference value between two sampling points is larger, then the credibility of the density difference factor between these two sampling points should be reduced; conversely, if the vegetation density at two sampling points is relatively consistent, then the vegetation condition difference value will be correspondingly smaller, and the credibility of the density difference factor between these two sampling points can be improved. Based on the foregoing logic, the vegetation condition difference value is used as the denominator part, and the density difference factor is used as the numerator part to obtain the density difference parameter between two sampling points.

[0042] It should be noted that the function of the preset first parameter is to prevent the denominator from being 0. Here, it can take the value of 0.001, and the specific value can be adjusted according to the actual situation and is not limited here.

[0043] Step S3: For any sampling point in the area to be measured, fuse and analyze the positional relationship between this sampling point and the remaining other sampling points, the differences in spectral data, and the density difference parameters at all sampling times to obtain the abnormal degree value of the building density at this sampling point; for any wavelength, fuse and analyze the fluctuations of the differences in the absorption rates of this wavelength at all sampling points at different times and the abnormal degree value of the building density at each sampling point to obtain the retention degree value of this wavelength.

[0044] Based on the method in Step S2, at any sampling time, the density difference parameter between any two sampling points at this sampling time can be obtained. To obtain the wavelength that can represent the density of the building, it is necessary to analyze the contribution degree of the spectral data at different positions to the building density. Moreover, in the hyperspectral data obtained by aerial photography, the differences in the spectral data of sampling points with close positions may be caused by the sparseness of the roads between buildings. Such differences cannot well show the difference relationship between the building density at this sampling point and the normal building density. Therefore, after obtaining the density difference parameter of the building between sampling points, determine the local range of each sampling point, consider the distance metric between the spectral data of sampling points within this local range, and combine the density difference parameter between sampling points to calculate the abnormal degree value of the building density at each sampling point, which is used to measure the contribution degree of the spectral data at this sampling point to the analysis of the building density.

[0045] Preferably, in an embodiment of the present invention, the method for obtaining the abnormal degree value of the building density includes: Please refer to Figure 3 , which shows the flowchart of a method for obtaining the abnormal degree value of the building density provided by an embodiment of the present invention. The method includes the following steps: Step S301: Determine the comparison point of each sampling point according to the positional relationship between sampling points.

[0046] By selecting the comparison point of each sampling point through the positional relationship between sampling points, it can be ensured that the sampling point and the selected comparison point are relatively close in space, thereby increasing the spatial representativeness of the data. Such a comparison point is more likely to have similar environmental conditions or influencing factors with the corresponding sampling point, increasing the accuracy of the subsequent process.

[0047] Optionally select a sampling point as the target point, calculate the Euclidean distance between the target point and the positions of the remaining each sampling point as the distance factor, and then use the sampling points with the distance factor less than the preset distance threshold as the comparison points of the target point.

[0048] It should be noted that the calculation formula of the Euclidean distance is a process well-known to those skilled in the art, and will not be elaborated here; the preset distance threshold is set to 7, and the specific value can be adjusted according to the implementation scenario, and will not be limited here; in other embodiments of the present invention, the value obtained by normalizing the Euclidean distance can also be used as the distance factor. At this time, the value of the distance factor is 0-1, and the closer it is to 1, the farther the distance between the two sampling points. The distance threshold is used to screen out the comparison points closer to the sampling point. Therefore, the preset distance threshold can be set to 0.5, and the smaller the value, the stricter the screening conditions for the comparison points, and vice versa, the looser.

[0049] Step S302: Determine the abnormal degree value of the building housing density at each sampling point according to the difference between the spectral data of each sampling point and its comparison point at all sampling times and the building housing density difference parameter.

[0050] For any sampling time, the difference in the absorption rate of the same wavelength in the spectral data of the target point and each comparison point is used as the absorption rate distance value, and then the sum value of the absorption rate distance values of all wavelengths is calculated as the distance sum value. This distance sum value can be regarded as the distance between the two spectral data. Then, according to the density difference parameter and the distance sum value between the target point and each comparison point, the deviation value of the spectral data of the target point and each comparison point at this sampling time is obtained, and the deviation value is negatively correlated with the density difference parameter, and the distance sum value and the deviation value are positively correlated. Among them, the positive correlation means that the dependent variable will increase as the independent variable increases, and the dependent variable will decrease as the independent variable decreases. The specific relationship can be a multiplicative relationship, an additive relationship, the power of an exponential function, etc., which is determined by the actual application; the negative correlation means that the dependent variable will decrease as the independent variable increases, and the dependent variable will increase as the independent variable decreases, which can be a subtractive relationship, a divisive relationship, etc., which is determined by the actual application.

[0051] The average value of the deviation values of the spectral data of the target point and each comparison point at all sampling times is used as the abnormal factor of the building housing density of the target point; the average value of the abnormal factors of the building housing density of the target point and all comparison points is used as the abnormal degree value of the building housing density at the target point. The target point takes the sampling point as an example, and the formula model of the abnormal degree value of the building housing density is: Among them, represents the abnormal degree value of the building housing density of the sampling point ; represents the total number of comparison points of the sampling point ; represents the total number of sampling times; represents at the th sampling time, the sampling point and the density difference parameter between the th and the It represents at the th sampling moment, the sum of distances between the spectral data of the sampling point and the th comparison point; It represents a preset second parameter.

[0052] In the formula model of the abnormal degree value of the building housing density, when the distance between the spectral data of the sampling point and its comparison point is large and the density difference parameter is small, the deviation value will be larger, indicating that the difference in spectral data between the two sampling points is more likely to be caused by the different degrees of density of the roads between the buildings at the two sampling points. Such data changes cannot show the density of the buildings at the sampling point, so the contribution degree is lower, and its abnormal degree is higher. Then, the mean value of the deviation values between the target point and its comparison points at all sampling moments is calculated as the abnormal factor of the building housing density at the target point . Finally, the mean value of the abnormal factors of the building housing density of the target point and all its corresponding comparison points is used as the abnormal degree value of the building housing density at the target point. The larger this value is, the lower the contribution degree of the spectral data of the sampling point to the analysis of the building housing density, that is, the higher the abnormal degree of the building housing density.

[0053] It should be noted that the function of the preset second parameter is to prevent the denominator from being 0. Here, it can take the value of 0.001, and the specific value can be adjusted according to the implementation scenario and is not limited here.

[0054] Thus, the abnormal degree value of the building housing density at each sampling point can be obtained, which is used to characterize the contribution degree of its spectral data to the building housing density.

[0055] Since the environmental factors change at different sampling moments, such as the weather changes, the water content in the air changes, resulting in the change of environmental humidity. This environmental change will affect the absorption rate of some wavelengths in the spectral data. At the same time, because the influence caused by the environmental change is relatively uniform, the change of the absorption rate of the wavelength caused by the environmental factor at different moments will be different from the change of the absorption rate of the wavelength caused by the building housing density situation. Therefore, based on this feature, combined with the abnormal degree value of the building housing density at the sampling point obtained above, the research value of each wavelength for studying the building housing density situation is quantified, that is, the retention degree value of each wavelength is obtained.

[0056] Preferably, in an embodiment of the present invention, the method for obtaining the retention degree value includes: Pair all the sampling moments to obtain all non-repeating moment combinations. For example, if there are 4 sampling moments, denoted as sampling moment 1, sampling moment 2, sampling moment 3, and sampling moment 4 respectively, then all the moment combinations include: (sampling moment 1, sampling moment 2), (sampling moment 1, sampling moment 3), (sampling moment 1, sampling moment 4), (sampling moment 2, sampling moment 3), (sampling moment 2, sampling moment 4), (sampling moment 3, sampling moment 4).

[0057] For any sampling point, in each moment combination, fuse the abnormal degree value of the building density at this sampling point with the difference in the absorption rate of the wavelength to be measured at this sampling point at two sampling moments: Take the value obtained by performing negative correlation mapping and normalization on the abnormal degree value of the building density at each sampling point as the retention weight; for any sampling point, take the difference in the absorption rate of the wavelength to be measured at this sampling point at two sampling moments as the absorption rate change factor; take the product of the absorption rate change factor at each moment combination of this sampling point and the corresponding retention weight as the absorption rate change characteristic value of this sampling point. Taking the wavelength as an example, the formula model of the absorption rate change characteristic value can be specifically, for example: Among them, represents the absorption rate change characteristic value of the wavelength to be measured at the sampling point in the th moment combination; represents the abnormal degree value of the building density at the sampling point ; represents at the th moment combination, the absorption rate of the wavelength to be measured at the sampling point at the first sampling moment; represents at the th moment combination, the absorption rate of the wavelength to be measured at the sampling point at the second sampling moment; represents the exponential function with the natural constant as the base.

[0058] In the formula model of the absorption rate change characteristic value, since the larger the abnormal degree value of the building density, it indicates that the contribution degree of the spectral data at this sampling point to the analysis of the building density is low, so the abnormal degree value of the building density at the sampling point is subjected to negative correlation mapping and normalization processing to achieve logical relationship correction and used as the retention weight ; at the same time, the absorption rate change factor It characterizes the differences in the absorption rates of the wavelength to be measured at the same sampling point at different times, and then combines the absorption rate change factor with the retention weight of the sampling point, taking the product of the two as the absorption rate change characteristic value of the sampling point at the wavelength to be measured. The absorption rate change characteristic value combines the contribution degree of the spectral data of the sampling point to the analysis of the building housing density and the change of the absorption rate of the wavelength to be measured at the same sampling point at different times. Therefore, analyzing the absorption rate change characteristic value in the subsequent process can better evaluate the retention situation of the wavelength to be measured.

[0059] Finally, the standard deviation of the absorption rate change characteristic values of all sampling points under the same time combination is used as the retention factor of the wavelength to be measured under the same time combination; the mean value of the retention factors of the wavelength to be measured under all sampling time combinations is used as the retention degree value of the wavelength to be measured. Taking the wavelength as an example, the formula model of the retention degree value can be specifically, for example: Among them, represents the retention degree value of the wavelength to be measured ; represents the total number of time combinations; represents the sampling point at the th time combination, and the absorption rate change characteristic value of the wavelength to be measured ; represents at the th time combination, the retention factor of the wavelength to be measured ; represents calculating the standard deviation.

[0060] Calculate the standard deviation of the absorption rate change characteristic values of the wavelength to be measured for all sampling points under the same time combination to obtain the retention factor of the wavelength to be measured. When the retention factor is larger, it indicates that the absorption rate changes of different sampling points, that is, at different positions, for the wavelength to be measured fluctuate greatly, then it is more likely that the state of the building itself has changed; on the contrary, if the retention factor is smaller, it indicates that the absorption rate changes of different sampling points, that is, at different positions, for the wavelength to be measured fluctuate less, indicating that the wavelength to be measured is more likely to be affected by environmental factors, so it has little reference value for studying the building housing density situation.

[0061] Based on the above process, the retention degree values of each wavelength can be obtained.

[0062] Step S4: Calculate the initial covariance matrix corresponding to the spectral data based on the hyperspectral data of all sampling points at all sampling times; correct the element values in the initial covariance matrix according to the retention degree values of each wavelength to obtain an updated covariance matrix; perform principal component analysis on the updated covariance matrix to obtain a dimensionality reduction result.

[0063] Principal component analysis is a commonly used dimensionality reduction technique mainly for discovering the most important patterns and relationships in data. In the embodiments of the present invention, principal component analysis is used to reduce the dimensionality of all spectral data, thereby extracting key features, and the dimensionality reduction result is used to reflect the density status of buildings. Constructing a covariance matrix is a necessary step and process in the principal component analysis process, which describes the linear relationship between variables. Therefore, an initial covariance matrix can be calculated first according to the hyperspectral data.

[0064] Preferably, in an embodiment of the present invention, the method for obtaining the initial covariance matrix includes: Since the hyperspectral data includes two position data and one spectral data; and the embodiments of the present invention mainly analyze and reduce the dimensionality of the spectral data, each spectral data is used as a column vector, and the length of the column vector is equal to the number of wavelengths, and the element value is the wavelength absorption rate.

[0065] Then, the column vectors corresponding to the spectral data of all sampling points at all sampling times are used to obtain an initial matrix, and the initial matrix is calculated to obtain an initial covariance matrix, and the number of rows and columns of the initial covariance matrix are both equal to the number of wavelengths. It should be noted that the process of calculating the covariance matrix in the principal component analysis method is well known to those skilled in the art and will not be elaborated here.

[0066] The initial covariance matrix can provide an overall view of the relationship between each wavelength in the dataset. Then, since the retention degree values of each wavelength are calculated in step S3, which reflects the importance degree of each wavelength, the initial covariance matrix can be corrected by the retention degree values of the wavelengths to obtain an updated covariance matrix, so as to emphasize the influence of important wavelengths and reduce the influence of unimportant wavelengths at the same time, which is beneficial to ensuring that the subsequent principal component analysis pays more attention to those wavelengths that are valuable for analyzing the density status of buildings.

[0067] Preferably, in an embodiment of the present invention, the method for obtaining the updated covariance matrix includes: Construct a correction matrix based on the retention degree values of each wavelength. The correction matrix is a diagonal matrix, and each element value on the diagonal is the retention degree value of each wavelength, and the element values in the remaining positions are 0.

[0068] Multiply the matrix obtained by multiplying the correction matrix by the initial covariance matrix by the correction matrix again to obtain the updated covariance matrix. The formula model of the updated covariance matrix is: Wherein, represents the updated covariance matrix; represents the correction matrix; Represents the initial covariance matrix.

[0069] In the formula model for updating the covariance matrix, multiplying the initial covariance matrix on the left by a correction matrix is equivalent to performing a row transformation on the initial covariance matrix, that is, scaling the element values in each row of the initial covariance matrix. And multiplying on the right by a correction matrix is equivalent to performing a column transformation, that is, scaling the element values in each column. The element values on the diagonal of the correction matrix are the retention degree values of each wavelength. Therefore, after the above processing, the importance weights of each wavelength can be added to the original dimensions of each wavelength, so that the subsequent dimensionality reduction results can better represent the density information of the building houses and reduce the influence degree of unimportant wavelengths.

[0070] It should be noted that the number of rows and columns of the correction matrix is equal to the total number of wavelengths. Taking the total number of wavelengths as 3 as an example, the correction matrix is a 3×3 matrix, and the element value in the first row and first column is the retention degree value of the first wavelength, the element value in the second row and second column is the retention degree value of the second wavelength, the element value in the third row and third column is the retention degree value of the third wavelength, and the element values in the remaining positions are 0.

[0071] After obtaining the updated covariance matrix, the relationship between each wavelength in the updated covariance matrix has been adjusted at this time. Therefore, the subsequent principal component analysis process can be continued for the updated covariance matrix to obtain the dimensionality reduction result. The obtained dimensionality reduction result can more accurately reflect the density information of the building houses because the influence of unimportant wavelengths has been eliminated, and it can provide a more accurate basis for urban management.

[0072] It should be noted that for the convenience of calculation, all the index data involved in the operations in the embodiments of the present invention have undergone data preprocessing to cancel the influence of the dimension. The specific means of canceling the dimension influence are well-known technical means to those skilled in the art and will not be limited here; the principal component analysis method used in the embodiments of the present invention is a well-known technical means to those skilled in the art, and the specific process will not be elaborated here.

[0073] In summary, the object of the embodiment of the present invention is to improve the importance degree of the wavelength reflecting the building houses, so as to improve the accuracy of the dimensionality reduction result representing the density status of the building houses in the subsequent dimensionality reduction process. First, hyperspectral data of the area to be measured at different sampling times is obtained. Since even at the same sampling time, if only the difference in wavelength absorption rates between sampling points is analyzed, it is impossible to distinguish well whether the difference is affected by the density of vegetation. Therefore, when analyzing the difference in the absorption rates of wavelengths between different sampling points and the fluctuation of the absorption rates of all sampling points at the same wavelength, the difference in the density of vegetation between different sampling points is combined therewith to obtain the density difference parameter between different sampling points at each sampling time. This parameter more accurately reflects the density difference between different sampling points under the influence of the density of vegetation. Further, in order to determine whether each wavelength determines the density of plants, it is necessary to analyze the contribution degree of the spectral data of different sampling points to the plant density. Since the difference in the spectral data of sampling points with similar positions may be caused by the different density of plants, based on the aforementioned density difference parameter, combined with the positional relationship between each sampling point, the abnormal degree value of the building house density at each sampling point is obtained, which is used to quantify the contribution degree of the spectral data at this sampling point to the building house density. Further, by performing a fusion analysis on the difference in the absorption rates of each wavelength at different sampling points at different times and the abnormal degree value of the building house density, and evaluating its fluctuation, the retention degree value of each wavelength can be obtained, which is used to distinguish the wavelengths with high research value and low research value for the density status of the building houses. Then, based on all the hyperspectral data, the initial covariance matrix is calculated, and the element values in the initial covariance matrix are corrected based on the retention degree values of each wavelength, so that the wavelengths with high retention degree values have higher weights, and the updated covariance matrix is obtained. Finally, the updated covariance matrix is continued to be subjected to principal component analysis, and the obtained dimensionality reduction result will contain more key information and can more accurately represent the density status of the building houses.

[0074] This embodiment also provides a building house remote sensing mapping data processing system, which includes a processor 500, a memory 501 and a computer program, wherein the memory 501 is used to store the corresponding computer program, and the processor 500 is used to run the corresponding computer program. When the computer program runs on the processor 500, it can implement the steps of any one of the building house remote sensing mapping data processing methods.

[0075] Please refer to Figure 4, which shows a system block diagram of a remote sensing mapping data processing system for building houses provided by an embodiment of the present invention, including: a data acquisition module 401: used to implement step S1 in the above method, a density difference analysis module 402: used to implement step S2 in the above method, a wavelength retention degree analysis module 403: used to implement step S3 in the above method, and a dimensionality reduction module 404: used to implement step S4 in the above method.

[0076] Please refer to Figure 5 , which shows a system schematic diagram provided by an embodiment of the present invention, including a processor 500, a memory 501, a bus 502, and a communication interface 503. The processor 500, the communication interface 503, and the memory 501 are connected through the bus 502; among them, the memory 501 may include a high-speed random access memory, and the bus 502 may be an ISA bus, a PCI bus, an EISA bus, etc. The processor 500 may be an integrated circuit chip with signal processing capabilities.

[0077] An embodiment of the present invention also provides a computer-readable storage medium corresponding to the method provided by the foregoing embodiment. Specifically, it may be, for example, an optical disc, on which a computer program (i.e., a program product) is stored. When the computer program is run by a processor, it will execute the method provided by any foregoing embodiment.

[0078] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), and other optical and magnetic storage media, which will not be elaborated here one by one.

[0079] It should be noted that: the above sequence of embodiments of the present invention is only for description and does not represent the advantages or disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0080] Each embodiment in this specification is described in a progressive manner. The same or similar parts among the embodiments can be referred to each other, and the key points of each embodiment are the differences from other embodiments.

Claims

1. A method for processing building remote sensing mapping data, characterized in that: The method comprises: Obtaining hyperspectral data of the area to be tested at different sampling times; For the hyperspectral data at any sampling time, the density difference parameters between different sampling points in the hyperspectral data at the sampling time are obtained according to the differences in vegetation density between different sampling points, the differences in the absorbance of different wavelengths at different sampling points, and the fluctuations in the absorbance of all sampling points at the same wavelength. For any sampling point in the area to be tested, the positional relationship between the sampling point and the remaining sampling points, the difference between the spectral data and the density difference parameters at all sampling times are fused and analyzed to obtain the abnormal density value of the building house at the sampling point; for any wavelength, the difference in the absorbance of the wavelength at all sampling points at different times and the abnormal density value of the building house at each sampling point are fused and analyzed for fluctuations to obtain the retention value of the wavelength; According to the hyperspectral data of all sampling points at all sampling times, the initial covariance matrix corresponding to the spectral data is calculated; the element values ​​in the initial covariance matrix are corrected according to the retention degree value of each wavelength to obtain an updated covariance matrix; the principal component analysis is performed on the updated covariance matrix to obtain a dimensionality reduction result.

2. A method for processing building remote sensing mapping data according to claim 1, characterized in that: The method for obtaining the density difference parameter includes: Obtain the normalized vegetation index at each sampling point in the area to be measured; At any sampling time, the difference in the normalized vegetation index between two different sampling points is taken as the vegetation condition difference value; The difference between the absorbances of two different sampling points at the same wavelength and the fluctuation of the absorbances of all sampling points at the same wavelength are analyzed to obtain the density difference factor between different sampling points at the sampling moment; According to the vegetation condition difference value and the density difference factor between two different sampling points, the density difference parameter between the two different sampling points at the sampling time is obtained, the density difference parameter is positively correlated with the density difference factor, the density difference parameter is negatively correlated with the vegetation condition difference value, and the value of the density difference parameter is a normalized value.

3. A method for processing building remote sensing mapping data according to claim 2, characterized in that: The method for obtaining the density difference factor includes: Select any wavelength as the wavelength to be measured, and at the sampling time, take the difference in absorbance between two different sampling points at the wavelength to be measured as the absorbance difference value; take the standard deviation of the absorbance of all sampling points at the wavelength to be measured as the absorbance fluctuation value; The absorbance change index of the two sampling points at the measured wavelength is obtained according to the absorbance fluctuation value and the absorbance difference value, and the absorbance fluctuation value and the absorbance difference value both change in the same direction as the absorbance change index; the average of the absorbance change index of the two different sampling points at all wavelengths is used as the density difference factor.

4. The method for processing building remote sensing mapping data according to claim 1, characterized in that: The method for obtaining the abnormality value of the housing building density includes: Select any sampling point as the target point, and select the comparison point of the target point from the remaining sampling points according to the distance between the target point and each remaining sampling point; For any sampling moment, the difference in the absorbance of the same wavelength in the spectral data of the target point and each comparison point is taken as the absorbance distance value; the sum of the absorbance distance values ​​of all wavelengths is taken as the distance sum value, and according to the density difference parameter between the target point and each comparison point and the distance sum value, the deviation value of the spectral data of the target point and each comparison point at the sampling moment is obtained, the deviation value is negatively correlated with the density difference parameter, and the distance sum value is positively correlated with the deviation value; The mean of the deviation values ​​of the spectral data of the target point and each comparison point at all sampling moments is taken as the building density anomaly factor of the target point; the mean of the building density anomaly factors of the target point and all comparison points is taken as the building density anomaly degree value at the target point.

5. A method for processing building remote sensing mapping data according to claim 4, characterized in that: The method for obtaining the comparison point includes: The Euclidean distance between the target point and each remaining sampling point is calculated as a distance factor, and the sampling points whose distance factors are less than a preset distance threshold are used as comparison points of the target point.

6. A method for processing building remote sensing mapping data according to claim 3, characterized in that: The method for obtaining the retention degree value comprises: Combine all sampling moments in pairs to obtain all non-repeating moment combinations; For any sampling point, in each time combination, the abnormality value of the building density at the sampling point is fused with the difference in the absorbance of the wavelength to be measured at the sampling point at two sampling moments to obtain the absorbance change characteristic value of the sampling point; The standard deviation of the absorbance change characteristic value of the measured wavelength under the same time combination of all sampling points is used as the retention factor of the measured wavelength under the same time combination; the mean of the retention factors of the measured wavelength under all time combinations is used as the retention degree value of the measured wavelength.

7. A method for processing building remote sensing mapping data according to claim 6, characterized in that: The method for obtaining the absorption rate change characteristic value comprises: The abnormality value of building density at each sampling point is negatively correlated and normalized, and the value is used as the retention weight; for any sampling point, in each time combination, the difference in the absorbance of the sampling point at the wavelength to be measured at two sampling times is used as the absorbance change factor; the product of the absorbance change factor of the sampling point at each time combination and the corresponding retention weight is used as the absorbance change characteristic value of the sampling point at the wavelength to be measured.

8. The method for processing building remote sensing mapping data according to claim 1, characterized in that: The method for obtaining the initial covariance matrix includes: The hyperspectral data includes two position data and one spectral data; Each spectral data is taken as a column vector, the length of the column vector is equal to the number of wavelengths, and the element value is the wavelength absorbance; The column vectors corresponding to the spectral data of all sampling points at all sampling moments form an initial matrix, and the initial matrix is ​​calculated to obtain an initial covariance matrix.

9. The method for processing building remote sensing mapping data according to claim 1, characterized in that: The method for obtaining the updated covariance matrix includes: A correction matrix is ​​constructed based on the retention degree values ​​of each wavelength, wherein the correction matrix is ​​a diagonal matrix, and each element value on the diagonal is the retention degree value of each wavelength, and the element values ​​at other positions are 0; The matrix obtained by multiplying the correction matrix by the initial covariance matrix is ​​then multiplied by the correction matrix to obtain the updated covariance matrix.

10. A building remote sensing mapping data processing system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 9 are implemented.

Images