Space heterodyne imaging spectral feature information extraction method based on principal component analysis

Abstract

The invention provides a spatial heterodyne imaging spectral feature information extraction method based on principal component analysis. The method comprises the following steps: acquiring original spectral data acquired by a spatial heterodyne interference spectral imager; preprocessing the original spectral data, wherein the preprocessing comprises the steps of denoising, normalization and the like; performing dimension reduction processing on the preprocessed spectral data by applying a principal component analysis (PCA) method, and extracting principal component features; constructing a spectral feature information extraction model according to the extracted principal component features; and performing feature extraction and classification identification on the new spectral data by using the constructed spectral feature information extraction model. According to the method, preprocessing and feature extraction are carried out on spectral data by using PCA, and an efficient spectral inversion model is constructed. Through the method, the precision and efficiency of spectrum inversion can be remarkably improved, the complexity and noise interference of data processing are reduced, and powerful support is provided for further application of a spatial heterodyne imaging spectrum technology.

Description

Technical Field

[0001] The present invention relates to the technical field of spectral imaging, and in particular to a method and a device for extracting spectral feature information of spatial heterodyne imaging based on principal component analysis (PCA), aiming to improve the resolution of spectral imaging and the accuracy of feature information extraction. Background Art

[0002] Spatial heterodyne interference spectral imaging technology is a new type of spectral analysis technology with advantages such as high resolution and wide spectral range. It is widely used in atmospheric detection, remote sensing measurement, material analysis and other fields. However, when processing complex spectral data, traditional spatial heterodyne imaging technology often faces problems such as large data volume and difficulty in extracting feature information. Principal component analysis (PCA), as an effective data dimension reduction and feature extraction method, can effectively solve these problems.

[0003] PCA is a commonly used data dimensionality reduction technique that converts possibly correlated variables into linearly uncorrelated variables (i.e., principal components) through orthogonal transformation. These principal components are sorted by variance, with the first principal component having the largest variance, i.e., containing the most information.

[0004] PCA can reveal the inherent structure in the data, reduce data redundancy and noise, and improve the efficiency and accuracy of data processing.

[0005] Spectral data usually contain a large amount of band information, and there may be correlation between the bands, which makes direct processing of spectral data complex and time-consuming.

[0006] By processing spectral data through PCA, the main characteristic components in the spectrum can be extracted, the data dimension can be reduced, and the amount of calculation can be reduced. At the same time, PCA can also enhance the useful information in the spectral signal, suppress noise and interference, and improve the accuracy and reliability of spectral information extraction.

[0007] In the large-aperture spatial heterodyne interference spectral imaging system, due to the increase in the field of view of the light, the interference pattern and the spectral information show a nonlinear relationship. This makes it difficult to directly apply the traditional Fourier transform method to the spectral inversion process, resulting in errors in the inversion results.

[0008] Faced with massive amounts of spectral data, traditional data processing methods are often inefficient and difficult to meet the needs of real-time processing. At the same time, additional noise and errors may be introduced during data processing, affecting the accuracy of spectral information extraction.

[0009] The present invention proposes a method for extracting feature information of spatial heterodyne imaging spectrum based on principal component analysis. The method uses PCA to preprocess and extract features of spectral data and construct an efficient spectral inversion model. Through this method, the accuracy and efficiency of spectral inversion can be significantly improved, the complexity of data processing and noise interference can be reduced, and strong support can be provided for the further application of spatial heterodyne imaging spectroscopy technology. Summary of the invention

[0010] The purpose of the present invention is to provide a method for extracting spectral feature information of spatial heterodyne imaging based on principal component analysis. The method realizes rapid dimensionality reduction and efficient feature extraction of spectral data by combining spatial heterodyne technology and principal component analysis, thereby improving the resolution and accuracy of spectral imaging.

[0011] To achieve the above objectives, the present invention adopts the following technical solutions.

[0012] A method for extracting spatial heterodyne imaging spectrum feature information based on principal component analysis includes the following steps.

[0013] A1 obtains the original spectral data collected by the spatial heterodyne interference spectral imager.

[0014] A2 preprocesses the raw spectral data, including denoising and normalization.

[0015] A3 applies the principal component analysis (PCA) method to reduce the dimension of the preprocessed spectral data and extract the principal component features.

[0016] A4 constructs a spectral feature information extraction model based on the extracted main component features.

[0017] A5 uses the constructed spectral feature information extraction model to perform feature extraction and classification identification on new spectral data.

[0018] The principal component analysis in A3 includes.

[0019] Compute the covariance matrix of the preprocessed spectral data.

[0020] Perform eigenvalue decomposition on the covariance matrix to obtain the eigenvalues ​​and corresponding eigenvectors.

[0021] Sort by the size of the eigenvalues ​​and select the eigenvectors corresponding to the first N largest eigenvalues ​​as the principal components.

[0022] The original spectral data are projected onto these principal components to obtain the spectral data after dimension reduction.

[0023] The spectral feature information extraction model in step A4 is constructed based on a machine learning algorithm, including but not limited to a support vector machine (SVM), a neural network, etc.

[0024] A device for extracting spatial heterodyne imaging spectrum feature information based on principal component analysis, comprising:

[0025] The data acquisition module is used to obtain the original spectral data of the spatial heterodyne interference spectral imager.

[0026] The data preprocessing module is used to perform preprocessing operations such as denoising and normalization on the original spectral data.

[0027] The principal component analysis module is used to perform principal component analysis on the preprocessed spectral data and extract the principal component features.

[0028] The feature extraction model building module is used to build a spectral feature information extraction model based on the extracted main component features.

[0029] The feature extraction and recognition module is used to extract features and classify and recognize new spectral data using the constructed spectral feature information extraction model.

[0030] The method and device for extracting spectral feature information of spatial heterodyne imaging based on principal component analysis of the present invention have the following beneficial effects: the resolution and accuracy of spectral imaging are improved, and rapid dimension reduction and efficient feature extraction of spectral data are achieved through principal component analysis; the data processing flow is simplified, the computational complexity is reduced, and the processing efficiency is improved; the interpretability of spectral data is enhanced, and more reliable technical support is provided for the application of spectral imaging technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 The present invention is a flowchart of a method for extracting spectral feature information of spatial outbound imaging based on principal component analysis.

[0032] Figure 2 It is a target interference data cube with interference information collected through spatial heterodyne imaging experiments.

[0033] Figure 3 The target spectrum data cube with spectral information is obtained by reconstructing the interference pattern of each pixel target point and performing Fourier transform.

[0034] Figure 4 These are the first 10 principal component images after the data cube is reduced in dimension using the atlas data cube compression algorithm based on principal component analysis. DETAILED DESCRIPTION

[0035] The present invention will be further described below with reference to specific examples.

[0036] The method for extracting spectral feature information of spatial outbound imaging based on principal component analysis described in this example includes the following steps.

[0037] A1 obtains the original spectral data collected by the spatial heterodyne interference spectral imager.

[0038] Furthermore, the specific process of A1 includes the following steps.

[0039] Spectral data collection is performed using a spatial heterodyne interferometer spectrometer. This instrument can capture the spectral information of the target scene in multiple bands and generate a high-dimensional raw spectral data set. These data usually contain rich spectral features, but may also be accompanied by noise and redundant information.

[0040] In the scanning imaging experiment, the displacement frequency of the displacement platform needs to be set to be consistent with the frequency of the spatial heterodyne imaging spectrometer to collect images. The spatial heterodyne imaging spectrometer moves one pixel each time to collect an interference image. The size of a pixel is 13um, and the displacement platform moves 13um at a time. In this experiment, the parameter value of the integration time is set to 255. It takes 6 seconds to collect each image, so the displacement time interval of the displacement platform is also set to 6 seconds. According to the principle of spatial heterodyne scanning imaging, the spatial heterodyne spectrometer moves one pixel each time and collects interference images, and a total of 2048 interference images are collected.

[0041] A2 preprocesses the raw spectral data, including denoising and normalization.

[0042] Furthermore, the specific process of A2 includes the following steps.

[0043] The original spectral data is denoised using appropriate denoising methods, such as the local mean and local standard deviation method, the spatial spectral decorrelation method (SSDC) or the optimized MNF (Maximum Noise Fraction) transformation method, etc. These methods can effectively reduce the noise in the image and improve the data quality.

[0044] The denoised spectral data is normalized to eliminate the dimensional differences between different variables and make the data in each band have the same magnitude. Normalization is usually achieved by subtracting the mean of each band and dividing it by its standard deviation.

[0045] A3 applies the principal component analysis (PCA) method to reduce the dimension of the preprocessed spectral data and extract the principal component features.

[0046] Furthermore, the specific process of A3 includes the following steps.

[0047] For the normalized spectral data, calculate its covariance matrix, which reflects the correlation between different bands.

[0048] Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and corresponding eigenvectors. The eigenvalues ​​indicate the importance of the corresponding eigenvectors, while the eigenvectors reflect the weight of the original data in the new principal component space.

[0049] According to the size of the eigenvalue, the first 10 principal components with larger eigenvalues ​​are selected. These principal components contain most of the information in the original data and reduce the dimension of the data.

[0050] The original spectral data are mapped into the new principal component space using the selected eigenvectors to obtain the reduced-dimensional data.

[0051] A4 constructs a spectral feature information extraction model based on the extracted main component features.

[0052] Based on the extracted principal component features, a spectral feature information extraction model is constructed. This model can effectively describe and represent spectral data using principal component features, providing a basis for subsequent feature extraction and classification recognition.

[0053] A5 uses the constructed spectral feature information extraction model to perform feature extraction and classification identification on new spectral data.

[0054] For new spectral data, the same preprocessing operation as step A2 is first performed, and then the spectral feature information extraction model constructed in step A4 is used to perform principal component analysis on the preprocessed data to extract the principal component features.

[0055] Based on the extracted principal component features, the spectral data is classified and identified using appropriate classification algorithms (such as support vector machines, neural networks, etc.). The classification and identification results can be used for subsequent applications, such as material identification, environmental monitoring, etc.

[0056] Through the above specific implementation methods, it is possible to effectively process and analyze the original spectral data collected by the spatial heterodyne interference spectroscopic imager, extract useful spectral feature information, and provide strong support for subsequent applications.

[0057] The specific implementation methods described above provide a detailed description of the technical solutions and beneficial effects of the present invention. It should be understood that the above are only effective embodiments of the present invention and are not intended to limit the present invention. Any modifications, supplements and equivalent substitutions made within the scope of the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for extracting spatial heterodyne imaging spectral feature information based on principal component analysis, comprising the following steps: A1 obtains the original spectral data collected by the spatial heterodyne interference spectral imager; A2 preprocesses the raw spectral data, including denoising and normalization steps; A3 applies the principal component analysis (PCA) method to reduce the dimension of the preprocessed spectral data and extract the principal component features; A4 constructs a spectral feature information extraction model based on the extracted main component features; A5 uses the constructed spectral feature information extraction model to perform feature extraction and classification identification on new spectral data.

2. The method according to claim 1, wherein the principal component analysis in step A3 comprises: Calculate the covariance matrix of the preprocessed spectral data; Perform eigenvalue decomposition on the covariance matrix to obtain the eigenvalues ​​and corresponding eigenvectors; Sort by eigenvalues ​​and select the eigenvectors corresponding to the first N largest eigenvalues ​​as principal components; The original spectral data are projected onto these principal components to obtain the spectral data after dimension reduction.

3. The method according to claim 1 or 2, wherein the spectral feature information extraction model in step A4 is constructed based on a machine learning algorithm, including but not limited to a support vector machine (SVM), a neural network, etc.

4. A spatial heterodyne imaging spectral feature information extraction device based on principal component analysis, comprising: A data acquisition module, used to obtain the original spectral data of the spatial heterodyne interference spectral imager; Data preprocessing module, used to perform preprocessing operations such as denoising and normalization on the original spectral data; The principal component analysis module is used to perform principal component analysis on the preprocessed spectral data and extract the principal component features; A feature extraction model building module is used to build a spectral feature information extraction model based on the extracted main component features; The feature extraction and recognition module is used to extract features and classify and recognize new spectral data using the constructed spectral feature information extraction model.

Images