Hyperspectral image feature extraction method based on end member extraction and MRPE algorithm

By using the normalized distance between the cells and end elements and the WMF and MRPE algorithms in hyperspectral images, the feature extraction accuracy problem caused by spectral uncertainty is solved, and the target detection accuracy is improved.

CN118941803APending Publication Date: 2024-11-12ROCKET FORCE UNIV OF ENG
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410988824.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-23
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

In hyperspectral images, when selecting the spectral neighbors of similarity measurements using Euclidean distance as a similarity measure function, the pixels belonging to the same land object are far away, reducing the target detection accuracy.

Method used

The normalized distance between the cell and the end element is used as a similarity measurement function, and combined with WMF and MRPE algorithms, the hyperspectral image is weighted mean filtered and feature extraction is performed to construct more accurate reconstruction weights to reduce spectral uncertainty. Influence.

Benefits of technology

The accuracy of feature data extraction is improved, so that the low-dimensional feature matrix more accurately reflects the essential features of the original image, and enhances the accuracy of object detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118941803A_ABST
    Figure CN118941803A_ABST
Patent Text Reader

Abstract

The invention discloses a hyperspectral image feature extraction method based on end member extraction and an MRPE algorithm, and the method comprises the following steps: 1, carrying out the weighted mean filtering processing of an original hyperspectral image through employing a computer through employing a WMF algorithm, and obtaining a denoised hyperspectral image; 2, performing matrix representation on the denoised hyperspectral image; 3, selecting a spectrum neighbor set by using the normalized distance between the pixel and the extracted end member; and 4, performing hyperspectral image feature extraction by adopting an MRPE algorithm according to the selected spectrum neighbor set. According to the method, the steps are simple, the design is reasonable, the normalized distance between the pixels and the end members is used for replacing the Euclidean distance to serve as a similarity measurement function between the pixels for spectrum neighbor selection, the influence of the spectrum uncertainty phenomenon can be effectively weakened, therefore, a more accurate reconstruction weight is constructed, and the feature data extraction accuracy is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of remote sensing image processing, and in particular relates to a hyperspectral image feature extraction method based on endmember extraction and MRPE algorithm. Background Art

[0002] Hyperspectral images have a spectral resolution of 10 -2 Spectral images within the range of λ. In the imaging process of hyperspectral images, due to the influence of factors such as noise, the spectral curves of the same type of objects often show differences, that is, the "same object, different spectrum" phenomenon. When the image dimension is high, this difference will have a serious impact on the similarity measurement between pixels, and thus bring great difficulties to subsequent use such as target detection or classification.

[0003] In hyperspectral images, pixels at adjacent positions are more likely to be composed of similar objects, and the closer the spatial distance, the greater the probability. Based on this feature, the spatial neighbor information of hyperspectral images can be used, but it uses Euclidean distance as the similarity measurement function between samples to select spectral neighbor pixels, which cannot effectively reduce the impact of spectral uncertainty, resulting in pixels belonging to similar objects in low-dimensional data being far away from each other, thereby reducing the accuracy of target detection.

[0004] Therefore, a hyperspectral image feature extraction method based on endmember extraction and MRPE algorithm is needed. The normalized distance between pixels and endmembers is used instead of Euclidean distance as the similarity measurement function between pixels for spectral neighbor selection, which can effectively reduce the influence of spectral uncertainty phenomenon, thereby constructing more accurate reconstruction weights, so that the low-dimensional feature matrix can more accurately reflect the essential characteristics of the original image and improve the accuracy of feature data extraction. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a hyperspectral image feature extraction method based on endmember extraction and MRPE algorithm in view of the deficiencies in the above-mentioned prior art. The method has simple steps and reasonable design. The normalized distance between pixels and endmembers is used instead of the Euclidean distance as the similarity measurement function between pixels for spectral neighbor selection, which can effectively reduce the influence of spectral uncertainty phenomenon, thereby constructing more accurate reconstruction weights, so that the low-dimensional feature matrix can more accurately reflect the essential characteristics of the original image and improve the accuracy of feature data extraction.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for extracting features of a hyperspectral image based on endmember extraction and MRPE algorithm, characterized in that the method comprises the following steps:

[0007] Step 1: Use a computer to perform weighted mean filtering on the original hyperspectral image using the WMF algorithm to obtain a denoised hyperspectral image;

[0008] Step 2: Matrix representation of the denoised hyperspectral image:

[0009] The denoised hyperspectral image is denoted as I by a computer, and the matrix of the denoised hyperspectral image I is represented as X, and X = [x1…x j …x n ], x j represents the data of the jth pixel in the denoised hyperspectral image I, j is a positive integer, and 1≤j≤n, and n represents the total number of pixels in the denoised hyperspectral image I; represents the data of the i-th band corresponding to the j-th pixel, d represents the total number of bands of the hyperspectral image I after denoising, i and d are both positive integers, and 1≤i≤d;

[0010] Step 3: Select the spectral neighbor set using the normalized distance between the pixel and the extracted endmember;

[0011] Step 4: Use the MRPE algorithm to extract hyperspectral image features based on the selected spectral neighbor set.

[0012] The above-mentioned method for extracting features of hyperspectral images based on endmember extraction and MRPE algorithm is characterized in that: Step three, the specific process is as follows:

[0013] Step 301: Using a computer, record the R×R square spectral neighbors around the j-th pixel as the neighbor space Ω(x j ), where R is a positive integer;

[0014] Step 302: Process the j-th pixel and other pixels in the denoised hyperspectral image I using a computer to obtain a normalized distance between the j-th pixel and other pixels;

[0015] Step 303: record the first k pixels with the smallest normalized distance in step 302 as the neighbor set K(x j ), where k is a positive integer.

[0016] The above-mentioned method for extracting features of hyperspectral images based on endmember extraction and MRPE algorithm is characterized in that: step 302, the specific process is as follows:

[0017] Step 3021: Use a computer to extract endmembers from the hyperspectral image using a vertex component analysis algorithm to obtain an endmember set E, where E = (e1…e b …e B ), where e brepresents the bth end member, B represents the total number of end members, b and B are positive integers, and 1≤b≤B; Represents the data of the i-th band in the b-th end member;

[0018] Step 3022: Using a computer according to the formula Get the spectral angle distance between the jth pixel and the bth end member;

[0019] Step 3023: Repeat step 3022 multiple times to obtain the spectral angle distance between each pixel and the bth end member, and obtain the minimum value of the spectral angle distances between all pixels and the bth end member. The maximum value of the spectral angle distances between all pixels and the bth end member

[0020] Step 3024: Use a computer to calculate the Get the normalized distance between the jth pixel and the pth pixel; where, represents the spectral angle distance between the p-th pixel and the b-th end member, p is a positive integer, and 1≤p≤n, p≠j;

[0021] Step 3025: record the first k pixels with the smallest normalized distance as the spectral neighbor set K(x j ), where k is a positive integer.

[0022] The above-mentioned method for extracting features of hyperspectral images based on endmember extraction and MRPE algorithm is characterized in that: Step 4, the specific process is as follows:

[0023] Step 401: Use a computer to use the MRPE algorithm to find the spectral neighbor set K(x j ) by minimizing the reconstruction error, obtaining the initial reconstruction weights, and comparing the jth pixel with the spectral neighbor set K(x j ) is denoted as w je ;in, e is a positive integer, and 1≤e≤E, E represents the spectral neighbor set K(x j )

[0024] Step 402: Use a computer to calculate the spectral neighbor set K(x j ) is judged by the e-th pixel. If the e-th pixel belongs to K(x j ) and belongs to the neighbor space Ω(x j ), then the reconstruction weight W is modified je =βw je ; If the e-th pixel belongs to K(x j ) but does not belong to the neighbor space Ω(x j), the modified reconstruction weight remains unchanged W je =w je ; Where β represents the correction parameter, and β≥1;

[0025] Step 403, repeating steps 401 to 402 multiple times to obtain a corrected reconstruction weight of each pixel;

[0026] Step 404: Using a computer and using the MRPE algorithm, input the corrected reconstruction weight of each pixel to obtain a projection matrix A;

[0027] Step 405: Use a computer to calculate the value of Y=A T X, and obtain the low-dimensional feature matrix Y of the hyperspectral image.

[0028] Compared with the prior art, the present invention has the following advantages:

[0029] 1. The method of the present invention has simple steps and reasonable design, and solves the problem that the current method of using Euclidean distance as a similarity measurement function between pixels to select spectral neighbor pixels cannot effectively reduce the influence of spectral uncertainty.

[0030] 2. The present invention uses the normalized distance between the pixel and the end member instead of the Euclidean distance as the similarity measurement function between pixels to perform spectral neighbor selection, which can effectively reduce the influence of spectral uncertainty, thereby constructing more accurate reconstruction weights, achieving more effective dimensionality reduction, and improving the accuracy of feature data extraction.

[0031] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 The figure is a flowchart of the method of the present invention. DETAILED DESCRIPTION

[0033] like Figure 1 As shown, the hyperspectral image feature extraction method based on endmember extraction and MRPE algorithm of the present invention comprises the following steps:

[0034] Step 1: Use a computer to perform weighted mean filtering on the original hyperspectral image using the WMF algorithm to obtain a denoised hyperspectral image;

[0035] Step 2: Matrix representation of the denoised hyperspectral image:

[0036] The denoised hyperspectral image is denoted as I by a computer, and the matrix of the denoised hyperspectral image I is represented as X, and X = [x1…x j …x n ], x jrepresents the data of the jth pixel in the denoised hyperspectral image I, j is a positive integer, and 1≤j≤n, and n represents the total number of pixels in the denoised hyperspectral image I; represents the data of the i-th band corresponding to the j-th pixel, d represents the total number of bands of the hyperspectral image I after denoising, i and d are both positive integers, and 1≤i≤d;

[0037] Step 3: Select the spectral neighbor set using the normalized distance between the pixel and the extracted endmember;

[0038] Step 4: Use the MRPE algorithm to extract hyperspectral image features based on the selected spectral neighbor set.

[0039] In this embodiment, step three, the specific process is as follows:

[0040] Step 301: Using a computer, record the R×R square spectral neighbors around the j-th pixel as the neighbor space Ω(x j ), where R is a positive integer;

[0041] Step 302: Process the j-th pixel and other pixels in the denoised hyperspectral image I using a computer to obtain a normalized distance between the j-th pixel and other pixels;

[0042] Step 303: record the first k pixels with the smallest normalized distance in step 302 as the neighbor set K(x j ), where k is a positive integer.

[0043] In this embodiment, step 302, the specific process is as follows:

[0044] Step 3021: Use a computer to extract endmembers from the hyperspectral image using a vertex component analysis algorithm to obtain an endmember set E, where E = (e1…e b …e B ), where e b represents the bth end member, B represents the total number of end members, b and B are positive integers, and 1≤b≤B; Represents the data of the i-th band in the b-th end member;

[0045] Step 3022: Using a computer according to the formula Get the spectral angle distance between the jth pixel and the bth end member;

[0046] Step 3023: Repeat step 3022 multiple times to obtain the spectral angle distance between each pixel and the bth end member, and obtain the minimum value of the spectral angle distances between all pixels and the bth end member. The maximum value of the spectral angle distances between all pixels and the bth end member

[0047] Step 3024: Use a computer to calculate the Get the normalized distance between the jth pixel and the pth pixel; where, represents the spectral angle distance between the p-th pixel and the b-th end member, p is a positive integer, and 1≤p≤n, p≠j;

[0048] Step 3025: record the first k pixels with the smallest normalized distance as the spectral neighbor set K(x j ), where k is a positive integer.

[0049] In this embodiment, step 4, the specific process is as follows:

[0050] Step 401: Use a computer to use the MRPE algorithm to find the spectral neighbor set K(x j ) by minimizing the reconstruction error, obtaining the initial reconstruction weights, and comparing the jth pixel with the spectral neighbor set K(x j ) is denoted as w je ;in, e is a positive integer, and 1≤e≤E, E represents the spectral neighbor set K(x j )

[0051] Step 402: Use a computer to calculate the spectral neighbor set K(x j ) is judged by the e-th pixel. If the e-th pixel belongs to K(x j ) and belongs to the neighbor space Ω(x j ), then the reconstruction weight W is modified je =βw je ; If the e-th pixel belongs to K(x j ) but does not belong to the neighbor space Ω(x j ), the modified reconstruction weight remains unchanged W je =w je ; Where β represents the correction parameter, and β≥1;

[0052] Step 403, repeating steps 401 to 402 multiple times to obtain a corrected reconstruction weight of each pixel;

[0053] Step 404: Using a computer and using the MRPE algorithm, input the corrected reconstruction weight of each pixel to obtain a projection matrix A;

[0054] Step 405: Use a computer to calculate the value of Y=A T X, and obtain the low-dimensional feature matrix Y of the hyperspectral image.

[0055] In this embodiment, R is a positive integer, and the value of R is 3;

[0056] In this embodiment, the WMF algorithm (weighted mean filter) is a weighted mean filter algorithm.

[0057] In this embodiment, weighted mean filtering is performed on the original hyperspectral image to eliminate the influence of noise points and reduce the spectral differences between similar ground objects.

[0058] In this embodiment, the MRPE algorithm, namely the Manifold Reconstruction Preserving Embedding algorithm, is a manifold reconstruction preserving embedding algorithm.

[0059] In this embodiment, k is 10 to 100, the size of the matrix X is d×n, the size of the projection matrix A is d×l, and the dimension l is 2 to 10;

[0060] In this embodiment, the AVIRIS data set, i.e., a hyperspectral data set, is used, the total number of bands d is selected as 196, and the total number of pixels n is 432. Through experiments of this method, for experiments on the AVIRIS data set, when B is 3, the algorithm effect is better. If the CEM target detection operator is used, the effect is better when β is 2; if the ACE or GLRT target detection operator is used, the effect is better when β is 3. In actual use, it can be adaptively adjusted.

[0061] In summary, the method of the present invention has simple steps and reasonable design. The normalized distance between the pixel and the end member is used instead of the Euclidean distance as the similarity measurement function between the pixels for spectral neighbor selection, which can effectively reduce the influence of the spectral uncertainty phenomenon, thereby constructing a more accurate reconstruction weight, so that the low-dimensional feature matrix can more accurately reflect the essential characteristics of the original image and improve the accuracy of feature data extraction.

[0062] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any simple modification, change and equivalent structural change made to the above embodiment based on the technical essence of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A hyperspectral image feature extraction method based on endmember extraction and MRPE algorithm, It is characterized in that The method comprises the following steps: Step 1: Use a computer to perform weighted mean filtering on the original hyperspectral image using the WMF algorithm to obtain a denoised hyperspectral image; Step 2: Matrix representation of the denoised hyperspectral image: The denoised hyperspectral image is recorded as I by computer, and the matrix of the denoised hyperspectral image I is expressed as X, and X = [x 1 …x j …x n ], x j represents the jth pixel in the denoised hyperspectral image I, j and n are positive integers, and 1≤j≤n, and n represents the total number of pixels in the denoised hyperspectral image I; represents the data of the i-th band corresponding to the j-th pixel, d represents the total number of bands of the hyperspectral image I after denoising, i and d are both positive integers, and 1≤i≤d; T represents transposition; Step 3: Select the spectral neighbor set using the normalized distance between the pixel and the extracted endmember; Step 4: Use the MRPE algorithm to extract hyperspectral image features based on the selected spectral neighbor set.

2. A method for extracting features from hyperspectral images based on endmember extraction and MRPE algorithm according to claim 1, Features: Step 3: The specific process is as follows: Step 301: Using a computer, record the R×R square spectral neighbors around the j-th pixel as the neighbor space Ω(x j ), where R is a positive integer; Step 302: Process the j-th pixel and other pixels in the denoised hyperspectral image I using a computer to obtain a normalized distance between the j-th pixel and other pixels; Step 303: record the first k pixels with the smallest normalized distance in step 302 as the neighbor set K(x j ), where k is a positive integer.

3. A method for extracting features from hyperspectral images based on endmember extraction and MRPE algorithm according to claim 2, Features: Step 302, the specific process is as follows: Step 3021: Use a computer to extract endmembers from the hyperspectral image using a vertex component analysis algorithm to obtain an endmember set E, where E = (e 1 …e b …e B ), where e b represents the bth end member, B represents the total number of end members, b and B are positive integers, and 1≤b≤B; Represents the data of the i-th band in the b-th end member; Step 3022: Using a computer according to the formula Get the spectral angle distance between the jth pixel and the bth end member; Step 3023: Repeat step 3022 multiple times to obtain the spectral angle distance between each pixel and the bth end member, and obtain the minimum value of the spectral angle distances between all pixels and the bth end member. The maximum value of the spectral angle distances between all pixels and the bth end member Step 3024: Use a computer to calculate the Get the normalized distance d(j,p) between the jth pixel and the pth pixel; where, represents the spectral angle distance between the p-th pixel and the b-th end member, p is a positive integer, and 1≤p≤n, p≠j; Step 3025: record the first k pixels with the smallest normalized distance as the spectral neighbor set K(x j ), where k is a positive integer.

4. A method for extracting features from hyperspectral images based on endmember extraction and MRPE algorithm according to claim 1, Features: Step 4: The specific process is as follows: Step 401: Use a computer to use the MRPE algorithm to find the spectral neighbor set K(x j ) by minimizing the reconstruction error, obtaining the initial reconstruction weights, and comparing the jth pixel with the spectral neighbor set K(x j ) is denoted as w je ;in, e is a positive integer, and 1≤e≤E, E represents the spectral neighbor set K(x j ) Step 402: Use a computer to calculate the spectral neighbor set K(x j ) is judged by the e-th pixel. If the e-th pixel belongs to K(x j ) and belongs to the neighbor space Ω(x j ), then the reconstruction weight W is modified je =βw je ; If the e-th pixel belongs to K(x j ) but does not belong to the neighbor space Ω(x j ), the modified reconstruction weight remains unchanged W je =w je ; Where β represents the correction parameter, and β≥1; Step 403, repeating steps 401 to 402 multiple times to obtain a corrected reconstruction weight of each pixel; Step 404: Using a computer and using the MRPE algorithm, input the corrected reconstruction weight of each pixel to obtain a projection matrix A; Step 405: Use a computer to calculate the value of Y=A T X, and obtain the low-dimensional feature matrix Y of the hyperspectral image.

Citation Information

Cited By

  • Hyperspectral end member extraction method and device combined with neighborhood constraint, equipment and medium

    CN122368792A