Hyperspectral Remote Sensing Image Classification Based on Local Reconstruction Fisher Analysis

Jiamin, 刘嘉敏 Liu; Song, 杨松 Yang; Hong, 黄鸿 Huang

doi:10.3788/cjl202047.0710001

Cited by 2 publications

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results show that the feature wavelength extracted by CARS algorithm is the best, and the correct rates of verification sets on ELM and SVM models are 98.75% and 100%, respectively. Hyperspectral technology has been widely used in quality testing, origin traceability and other fields because of its advantages of unified map, high speed and no pollution [11,12]. Hyperspectral technology has been widely used in quality testing, origin tracing and other fields because of its advantages such as unified map, high speed, no pollution and so on.…”

Section: Related Workmentioning

confidence: 99%

Nondestructive Identification of Cyperus Esculentus Based on Machine Learning and Vis-NIR Hyperspectral Information

2022

Proceedings of 2022 the 12th International Workshop on Computer Science and Engineering

View full text Add to dashboard Cite

At present, the detection of Cyperus esculentus is limited to manual testing and testing based on physical and chemical properties, which is time-consuming and damaging, so it is of great significance to establish a non-destructive testing model for the classification of Cyperus esculentus on the market. Vis-NIR hyperspectral images were used to identify three kinds of Cyperus esculentus from different producing areas. The Vis-NIR Hyperspectral system (382.19nm~1026.66 nm) was used to collect 600 samples and extract the region of interest (ROI). Taking the ROI of a single seed as a sample, a total of 600 sample data were obtained. Firstly, the Savitzky-Golay (SG) algorithm is used to preprocess the sample data. Then Competitive Adapative Reweighted Sampling (CARS) and Successive Projections Algorithm (SPA) are used to reduce the dimensionality of the preprocessed data. Finally, the reduced-dimension features are input into Support Vector Machine (SVM), Extreme Learning Machine (ELM) and Softmax classifier for training. The results show that the reduced-dimensional data can greatly reduce the data redundancy and then reduce the complexity of the model. The experimental results show that the verification accuracy based on SVM, ELM and Softmax model is up to 85.1%, which can effectively achieve rapid non-destructive testing.

show abstract

Section: Related Workmentioning

confidence: 99%

Nondestructive Identification of Cyperus Esculentus Based on Machine Learning and Vis-NIR Hyperspectral Information

2022

Proceedings of 2022 the 12th International Workshop on Computer Science and Engineering

View full text Add to dashboard Cite

show abstract

融合字典学习与视觉转换器的高分遥感影像场景分类方法

He Xiaojun,

Liu Xuan,

Wei Xian

2023

激光与光电子学进展

View full text Add to dashboard Cite

Classification methods of remote sensing scene images are mostly based on traditional machine learning or convolutional neural networks. The feature extraction capability of such methods is extremely limited, particularly for optical remote sensing images with large interclass similarity, complex spatial information, and various geometric structures, there are problems such as loss of feature information and low classification accuracy. To overcome these problems, we propose a highresolution remote sensing scene image classification method that combines dictionary learning and Vision Transformer (ViT). This method can not only mine the longdistance dependencies inside the images but can also use dictionary learning to capture the deep nonlinear structural information of images to improve classification accuracy. Through extensive experiments performed on the RSSCN7, NWPU -RESISC45, and Aerial Image Data Set (AID) public remote sensing image datasets trained from scratch on the PyTorch deep learning framework, the effectiveness of the proposed method is verified; the results show that the classification accuracy of the proposed method for the mentioned datasets is 1. 763 percentage points, 1. 321 percentage points, and 3. 704 percentage points higher than that of the original visual converter model, respectively. Moreover, the proposed method outperforms other advanced scene classification methods.

show abstract

Hyperspectral Remote Sensing Image Classification Based on Local Reconstruction Fisher Analysis

Cited by 2 publications

References 0 publications

Nondestructive Identification of Cyperus Esculentus Based on Machine Learning and Vis-NIR Hyperspectral Information

Nondestructive Identification of Cyperus Esculentus Based on Machine Learning and Vis-NIR Hyperspectral Information

融合字典学习与视觉转换器的高分遥感影像场景分类方法

Contact Info

Product

Resources

About