Novel Spectral Loss Function for Unsupervised Hyperspectral Image Segmentation

Article Geochemistry & Geophysics

Dual-View Spectral and Global Spatial Feature Fusion Network for Hyperspectral Image Classification

Tan Guo et al.

Summary: In this article, a dual-view spectral and global spatial feature fusion network (DSGSF) is proposed for hyperspectral image (HSI) classification. The DSGSF consists of a spatial subnetwork and a spectral subnetwork, which can extract spatial-spectral features with strong discriminating performance.

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING (2023)

Add to Collection

Article Geochemistry & Geophysics

Hyperspectral Remote Sensing Benchmark Database for Oil Spill Detection With an Isolation Forest-Guided Unsupervised Detector

Puhong Duan et al.

Summary: Oil spill detection is a topic of increasing importance due to its negative impact on the environment and coastal communities. This study proposes an unsupervised method for oil spill detection using hyperspectral remote sensing images. The method utilizes isolation forest to estimate the probability of each pixel belonging to seawater or oil spills, and generates pseudolabeled training samples using clustering algorithms. The proposed method achieves superior detection performance compared to other state-of-the-art approaches.

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING (2023)

Add to Collection

Article Geochemistry & Geophysics

Classification via Structure-Preserved Hypergraph Convolution Network for Hyperspectral Image

Yule Duan et al.

Summary: This paper proposes a structure-preserved hypergraph convolutional network (SPHGCN) for hyperspectral image classification. It introduces a multiple neighborhood reconstruction model to reveal the essential resemblance of multiple pixels in the nonlinear spectral feature space. The SPHGCN design utilizes a hypergraph convolution operation and a structure preservation layer to achieve more discriminative features from multiple pixel nodes. Experimental results demonstrate that the SPHGCN outperforms several state-of-the-art methods for hyperspectral image classification.

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING (2023)

Add to Collection

Article Geochemistry & Geophysics

Efficient Semantic Segmentation of Hyperspectral Images Using Adaptable Rectangular Convolution

Jose L. Garcia et al.

Summary: This letter presents a new technique based on adaptable rectangular convolutions (ARCs) to reduce computational cost and improve model accuracy for better results in the context of HSI semantic segmentation.

IEEE GEOSCIENCE AND REMOTE SENSING LETTERS (2022)

Add to Collection

Article Geochemistry & Geophysics

A Novel Transformer Based Semantic Segmentation Scheme for Fine-Resolution Remote Sensing Images

Libo Wang et al.

Summary: The fully convolutional network (FCN) with an encoder-decoder architecture is widely used for semantic segmentation. In this paper, the authors propose using the Swin Transformer as the backbone and a novel decoder called DCFAM for better context extraction and resolution restoration. Experimental results on two remotely sensed semantic segmentation datasets demonstrate the effectiveness of the proposed scheme.

IEEE GEOSCIENCE AND REMOTE SENSING LETTERS (2022)

Add to Collection

Article Environmental Sciences

Land cover classification through Convolutional Neur-al Network model assembly: A case study of a local rural area in Thailand

D. Fitton et al.

Summary: This paper proposes a methodology to build a land cover by aggregating the prediction of different CNN models, achieving high accuracy in predicting different land cover classes. The approach is demonstrated on a rural area in Thailand, showing improved accuracy in road classification.

REMOTE SENSING APPLICATIONS-SOCIETY AND ENVIRONMENT (2022)

Add to Collection

Article Geochemistry & Geophysics

Unsupervised Feature Learning Using Recurrent Neural Nets for Segmenting Hyperspectral Images

Lukasz Tulczyjew et al.

Summary: The paper introduces an unsupervised feature learning technique for hyperspectral image analysis, extracting discriminative features from data by learning compressed representations. The experiments show that this method achieves high-quality segmentation at a faster speed, with the advantage of data reduction.

IEEE GEOSCIENCE AND REMOTE SENSING LETTERS (2021)

Add to Collection

Article Geochemistry & Geophysics

Fusion of Dual Spatial Information for Hyperspectral Image Classification

Puhong Duan et al.

Summary: The research proposes a novel framework for hyperspectral image classification, focusing on extracting spatial information. By fusing dual spatial information, the framework not only accurately extracts discriminative features from HSIs, but also achieves better classification performance through spatial optimization.

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING (2021)

Add to Collection

Article Geochemistry & Geophysics