CN111695636B - Hyperspectral image classification method based on graph neural network - Google Patents
Hyperspectral image classification method based on graph neural network Download PDFInfo
- Publication number
- CN111695636B CN111695636B CN202010543653.XA CN202010543653A CN111695636B CN 111695636 B CN111695636 B CN 111695636B CN 202010543653 A CN202010543653 A CN 202010543653A CN 111695636 B CN111695636 B CN 111695636B
- Authority
- CN
- China
- Prior art keywords
- graph
- hyperspectral image
- neural network
- data
- super
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
- G06F18/2414—Smoothing the distance, e.g. radial basis function networks [RBFN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Abstract
The invention discloses a hyperspectral image classification method based on a graph neural network, which comprises the following steps: s1: acquiring hyperspectral image data of a region to be processed, and preprocessing the hyperspectral image data; s2: performing super-pixel segmentation on the preprocessed hyperspectral image data to obtain segmented super-pixels; s3: constructing graph data by the super pixels and constructing edges by adjacent super pixels; s4: training the graph data by utilizing a graph neural network model to obtain a classification result; the invention improves the precision and effect of the final classification result.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a hyperspectral image classification method based on a graph neural network.
Background
At present, with the development of remote sensing technology and imaging spectrometers, the resolution of hyperspectral images is continuously improved, the application requirements are more and more extensive, but the hyperspectral images have the characteristics of multiple wave bands, huge data volume and the like, and great difficulties are brought to classification, identification and the like of hyperspectral images.
However, because of the excessively high dimensions and highly redundant information, a dramatic increase in computational complexity may result and may affect the accuracy of classification. Meanwhile, a lot of manpower and material resources are required for obtaining the label sample of the hyperspectral data, and the label sample is very difficult to obtain, so that research on hyperspectral image classification is also hindered. The purpose of the study of hyperspectral image classification is to assign a feature class to each pixel in the hyperspectral image by means of the spectral information of one pixel, to extract features manually at the beginning, and then to model and classify the data by means of machine learning algorithms, since the manual extraction of features by researchers is required, the experiment is slow and the results are unstable.
Therefore, how to provide a hyperspectral image classification method capable of solving the above problems is a problem that a person skilled in the art needs to solve.
Disclosure of Invention
In view of the above, the invention provides a hyperspectral image classification method based on a graph neural network, reduces the dependence on a marked sample, provides a semi-supervised hyperspectral image classification algorithm based on the graph neural network, improves the precision of a final classification result by using a small amount of label data, reduces the occurrence of bad boundary contours, and obtains a classification result with high usability.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
a hyperspectral image classification method based on a graph neural network comprises the following steps:
s1: acquiring hyperspectral image data of a region to be processed, and preprocessing the hyperspectral image data;
s2: performing super-pixel segmentation on the preprocessed hyperspectral image data to obtain segmented super-pixels;
s3: forming edges by the adjacent super pixel points, and constructing graph data by the edges;
s4: and training the graph data by utilizing a graph network model to obtain a classification result.
The beneficial effects of the method are as follows: the HSGACN network model, namely the graph neural network, is introduced into hyperspectral image classification, so that the hyperspectral image classification effect and accuracy are improved.
Preferably, the step S1 specifically includes: the preprocessing process is to remove interference wave bands from the hyperspectral image data and normalize the removed hyperspectral image data.
Preferably, in the step S3, a spatial distance and a spectral distance between the two super pixels are obtained, and the spatial distance and the spectral distance are iterated after balancing weights.
Preferably, the number of iterations is 10.
Preferably, the graph neural network model is a network designed based on various graph neural network models, such as graph convolution network, graph attention network and the like.
Compared with the prior art, the hyperspectral image classification method based on the graph neural network reduces dependence on a marked sample, and provides a semi-supervised hyperspectral image classification algorithm based on the graph neural network, wherein the precision of a final classification result is improved by using a small amount of label data, and poor boundary contours are reduced to obtain the classification result with high usability.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present invention, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a fusion method of oblique photography models provided by the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
Referring to fig. 1, embodiment 1 of the invention discloses a hyperspectral image classification method based on a graph neural network, which comprises the following steps:
s1: acquiring hyperspectral image data of a region to be processed, and preprocessing the hyperspectral image data;
s2: performing super-pixel segmentation on the preprocessed hyperspectral image data to obtain segmented super-pixels;
step S2 provides a hyperspectral image super-pixel segmentation algorithm based on an SLIC algorithm, and the hyperspectral image data formed by super-pixels is finally obtained by calculating the spatial distance and the spectral distance between pixel points, balancing weights, iteratively updating the super-pixel cluster center and the range boundary, and stopping iteration when the error between the new cluster center and the old cluster center is smaller than a certain range.
S3: forming edges by the adjacent super pixel points, and constructing graph data by the edges;
wherein, in the specific process of step S3: the super-pixels in the segmentation graph are formed into individual nodes, the characteristics of the nodes are the average value of all pixel characteristics in the super-pixels, and then the nodes of adjacent super-pixels are formed into an edge to construct graph data.
S4: training the graph data by utilizing a graph neural network model to obtain a classification result.
The characteristics of unlabeled samples are changed by utilizing the characteristic propagation characteristics of the graph neural network, so that the characteristics of super pixels of the same class are more similar, the structure of graph data is gradually optimized by utilizing the characteristic that the graph attention network can change the weight of edges, and the characteristic of global operation can be realized by utilizing the graph convolution network, so that the characteristic and the structural information of the graph data are fully utilized.
Specifically, the graph neural network model in step S4 is a two-layer graph neural network structure, one layer is a graph annotation layer, and the other layer is a graph convolution layer, wherein the graph convolution network is an existing structure
In a specific embodiment, step S1 specifically includes: the preprocessing process is to remove interference wave bands from hyperspectral image data and normalize the removed hyperspectral image data.
Specifically, the interference wave band mainly eliminates whether the wave band is the absorption wave band (4-7 microns) of water vapor, and as the wave band of 4-7 microns is the interference absorption of satellite distance far atmospheric water, the data quality of the wave band is poor, the subsequent calculation is difficult, and the normalization is performed through mean square difference normalization (x-average value)/variance.
In a specific embodiment, in step S3, the spatial distance and the spectral distance between the two super pixels are obtained, and the spatial distance and the spectral distance are iterated after balancing weights.
Specifically, the expression of the spatial distance and the spectral distance is:
where dc is the spectral square difference, d (sx, sy) is the spectral angular distance, ds is the distance square difference, and m is typically 100 by adjusting m to balance.
In a specific embodiment, the number of iterations is 10, and the error expression between the cluster center and the old cluster center is
When the error is less than 0.01.
To verify the effectiveness of the method, experiments were performed on three public data sets Indian pins, pavia University and Kennedy Space Center to an accuracy of 91.84%, 95.69% and 98.42% respectively with only 30 marker samples per category.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (5)
1. The hyperspectral image classification method based on the graph neural network is characterized by comprising the following steps of: s1: acquiring hyperspectral image data of a region to be processed, and preprocessing the hyperspectral image data; s2: performing super-pixel segmentation on the preprocessed hyperspectral image data to obtain segmented super-pixels; s3: forming edges by the adjacent super pixel points, and constructing graph data by the edges; s4: training the graph data by utilizing a graph neural network model to obtain a classification result; wherein, in the specific process of step S3: forming super pixels in the segmentation graph into nodes, wherein the characteristics of the nodes are the average value of all pixel characteristics in the super pixels, and then constructing an edge of the nodes of adjacent super pixels to construct graph data; in step S4, the graph neural network model is a two-layer graph neural network structure, one layer is a graph annotation layer, the weight of the edge is changed to optimize the structure of the graph data, and the other layer is a graph convolution layer, so that the characteristics and the structural information of the graph data are fully utilized for global operation.
2. The hyperspectral image classification method based on the graph neural network according to claim 1, wherein the step S1 specifically includes: the preprocessing process is to remove interference wave bands from the hyperspectral image data and normalize the removed hyperspectral image data.
3. The hyperspectral image classification method based on the graph neural network according to claim 1, wherein in the step S3, a spatial distance and a spectral distance between two super pixels are obtained, and the spatial distance and the spectral distance balance weight are iterated after updating the clustering center position and the clustering range.
4. The hyperspectral image classification method based on the graph neural network according to claim 1, wherein in the step S4, the graph neural network model is a network designed by a graph roll-up network and a graph annotation network.
5. A hyperspectral image classification method based on a graph neural network as claimed in claim 3 wherein the number of iterations is 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010543653.XA CN111695636B (en) | 2020-06-15 | 2020-06-15 | Hyperspectral image classification method based on graph neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010543653.XA CN111695636B (en) | 2020-06-15 | 2020-06-15 | Hyperspectral image classification method based on graph neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111695636A CN111695636A (en) | 2020-09-22 |
| CN111695636B true CN111695636B (en) | 2023-07-14 |
Family
ID=72481169
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010543653.XA Active CN111695636B (en) | 2020-06-15 | 2020-06-15 | Hyperspectral image classification method based on graph neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111695636B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112329818B (en) * | 2020-10-20 | 2023-07-07 | 南京信息工程大学 | Hyperspectral image non-supervision classification method based on graph convolution network embedded characterization |
| CN112949416B (en) * | 2021-02-04 | 2022-10-04 | 东华理工大学 | Supervised hyperspectral multiscale graph volume integral classification method |
| CN113052216B (en) * | 2021-03-15 | 2022-04-22 | 中国石油大学(华东) | Oil spill hyperspectral image detection method based on two-way graph U-NET convolutional network |
| CN113239938B (en) * | 2021-05-11 | 2024-01-09 | 中国人民解放军火箭军工程大学 | Hyperspectral classification method and hyperspectral classification system based on graph structure |
| CN113269196B (en) * | 2021-07-19 | 2021-09-28 | 湖南大学 | Method for realizing hyperspectral medical component analysis of graph convolution neural network |
| CN113723255B (en) * | 2021-08-24 | 2023-09-01 | 中国地质大学（武汉） | Hyperspectral image classification method and storage medium |
| CN114399002A (en) * | 2022-01-21 | 2022-04-26 | 平安科技（深圳）有限公司 | Image scene classification method and device based on graph neural network |
| CN116109522B (en) * | 2023-04-10 | 2023-07-14 | 北京飞渡科技股份有限公司 | Contour correction method, device, medium and equipment based on graph neural network |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106469316A (en) * | 2016-09-07 | 2017-03-01 | 深圳大学 | The sorting technique of the high spectrum image based on super-pixel level information fusion and system |
| CN110619370A (en) * | 2019-09-23 | 2019-12-27 | 云南电网有限责任公司电力科学研究院 | Hyperspectral image super-pixel local linear embedding dimension reduction method |
| CN111160478A (en) * | 2019-12-31 | 2020-05-15 | 北京理工大学重庆创新中心 | Hyperspectral target significance detection method based on deep learning |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9430499B2 (en) * | 2014-02-18 | 2016-08-30 | Environmental Systems Research Institute, Inc. | Automated feature extraction from imagery |
| WO2019178702A1 (en) * | 2018-03-23 | 2019-09-26 | The Governing Council Of The University Of Toronto | Systems and methods for polygon object annotation and a method of training an object annotation system |
-
2020
- 2020-06-15 CN CN202010543653.XA patent/CN111695636B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106469316A (en) * | 2016-09-07 | 2017-03-01 | 深圳大学 | The sorting technique of the high spectrum image based on super-pixel level information fusion and system |
| WO2018045626A1 (en) * | 2016-09-07 | 2018-03-15 | 深圳大学 | Super-pixel level information fusion-based hyperspectral image classification method and system |
| CN110619370A (en) * | 2019-09-23 | 2019-12-27 | 云南电网有限责任公司电力科学研究院 | Hyperspectral image super-pixel local linear embedding dimension reduction method |
| CN111160478A (en) * | 2019-12-31 | 2020-05-15 | 北京理工大学重庆创新中心 | Hyperspectral target significance detection method based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| Geologic Body Classification of Hyperspectral Data Based on Dilated Convolution Neural Network at Tianshan Area;Jin Qin等;《 IGARSS 2019 - 2019 IEEE International Geoscience and Remote Sensing Symposium》;全文 * |
| 面向高光谱图像分类的超像素级Gabor特征融合方法研究;贾森等;《南京信息工程大学学报(自然科学版)》(01);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111695636A (en) | 2020-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111695636B (en) | Hyperspectral image classification method based on graph neural network | |
| WO2021003951A1 (en) | Hyperspectral image classification method based on label-constrained elastic network graph model | |
| Chen et al. | Single image shadow detection and removal based on feature fusion and multiple dictionary learning | |
| Jia et al. | Detection of cervical cancer cells based on strong feature CNN-SVM network | |
| CN107622104B (en) | Character image identification and marking method and system | |
| WO2020200030A1 (en) | Neural network training method, image processing method, image processing device, and storage medium | |
| Montazer et al. | An improved radial basis function neural network for object image retrieval | |
| CN111753828B (en) | Natural scene horizontal character detection method based on deep convolutional neural network | |
| CN110163239B (en) | Weak supervision image semantic segmentation method based on super-pixel and conditional random field | |
| CN112287839B (en) | SSD infrared image pedestrian detection method based on transfer learning | |
| CN109886161B (en) | Road traffic identification recognition method based on likelihood clustering and convolutional neural network | |
| Ghorai et al. | Multiple pyramids based image inpainting using local patch statistics and steering kernel feature | |
| CN109033978B (en) | Error correction strategy-based CNN-SVM hybrid model gesture recognition method | |
| CN110990775B (en) | Multi-view clustering method based on multi-manifold dual graph regularized non-negative matrix factorization | |
| CN109190511B (en) | Hyperspectral classification method based on local and structural constraint low-rank representation | |
| CN114119803A (en) | Scene image generation method based on causal graph | |
| CN114170608A (en) | Super-resolution text image recognition method, device, equipment and storage medium | |
| CN108090913B (en) | Image semantic segmentation method based on object-level Gauss-Markov random field | |
| CN102436645B (en) | Spectral clustering image segmentation method based on MOD dictionary learning sampling | |
| CN111126169A (en) | Face recognition method and system based on orthogonalization graph regular nonnegative matrix decomposition | |
| CN112329818B (en) | Hyperspectral image non-supervision classification method based on graph convolution network embedded characterization | |
| CN112232119A (en) | Remote sensing texture image segmentation method and device | |
| CN109190645B (en) | High-order high-dimensional image data representation and classification method | |
| CN114283431B (en) | Text detection method based on differentiable binarization | |
| CN116630610A (en) | ROI region extraction method based on semantic segmentation model and conditional random field |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |