Extracting spatial and spectral features through deep neural networks has become an effective means of classification hyperspectral images. However, most rarely consider the extraction multi-scale cannot fully integrate features. In order to solve these problems, this paper proposes a multi-level spectral-spatial feature fusion network (MSSN) for image classification. The uses original 3D cube as input data does not need use engineering. MSSN, using different scale neighborhood blocks...

Convolutional neural networks (CNNs) can extract advanced features of joint spectral–spatial information, which are useful for hyperspectral image (HSI) classification. However, the patch-based neighborhoods samples with fixed sizes usually used as input CNNs, cannot dig out homogeneousness between pixels within and outside patch. In addition, spatial quite different in spectral bands, not fully utilized by existing methods. this paper, a two-branch convolutional network based on...

In Hyperspectral image (HSI) classification, combining spectral information with spatial has become an efficient measure to obtain good classification results, where is generally introduced in unsupervised way or some complicated way. We introduce coordinates as the a simple supervised and propose two HSI algorithms, of samples are regarded features samples. A spectral-spatial algorithm proposed, named Classification Based on Spectral-Spatial Feature Fusion using Spatial Coordinates...

To achieve effective deep fusion features for improving the classification accuracy of hyperspectral remote sensing images (HRSIs), a pixel frequency spectrum feature is presented and introduced to convolutional neural networks (CNNs). Firstly, fast Fourier transform performed on each spectral obtain amplitude spectrum, i.e., feature. Then, obtained combined with form mixed feature, (SFMF). Several multi-branch CNNs fed SFMF, pixel, spatial are designed extracting features. A pre-learning...

In the field of hyperspectral image (HSI) classification in remote sensing, combination spectral and spatial features has gained considerable attention. addition, multiscale feature extraction approach is very effective at improving accuracy for HSIs, capable capturing a large amount intrinsic information. However, some existing methods extracting can only generate low-level consider limited scales, leading to low results, dense-connection based enhance propagation cost high model...

Although the deep-learning method has achieved great success for hyperspectral image (HSI) classification, few-shot HSI classification deserves sufficient study because it is difficult and expensive to acquire labeled samples. In fact, meta-learning methods can improve performance effectively. However, most of existing are supervised, which still heavily rely on data meta-training. Moreover, there many cross-scene tasks in real world, domain adaptation unsupervised been ignored so far. To...

Many systems in social world can be represented by complex networks. It is of great significance to detect the community structure and analyze functions for In recent years, plenty research works have been focused on this problem. paper, we propose an enhanced algorithm based ant colony optimization (ACO) detection problems. order avoid redundant computing ACO, divide into two groups, original group intelligent group, which search solution space simultaneously. due locus-based adjacency...

Convolutional neural networks (CNNs) have been widely used in hyperspectral image (HSI) classification. Many algorithms focus on the deep extraction of a single kind feature to improve There few studies two or more kinds fusion features and combination spatial spectral for The authors this paper propose an HSI spectral–spatial classification method based adaptive (SSDF). This first implements features, then it performs fused features. In SSDF, U-shaped network model with principal component...

Dynamic multi-objective optimization problem (DMOP) is such a type of problems that multiple contradictory objectives change over time. This paper designs special point-based hybrid prediction strategy (SHPS) integrated into the decomposition-based algorithm with differential evolution (MOEA/D-DE) to handle DMOPs, which denoted as MOEA/D-DE-SHPS. In SHPS, when historical information insufficient establish model population (PPS), (PRE) and variation (VAR) method are adapted generate initial...

Noise in hyperspectral images (HSIs) may degrade the HSI classification result. Robust principal component analysis (RPCA) is an excellent method to obtain low-rank (LR) representation of data and widely used image denoising also classification. However, are drawn as a union from multiple subspaces HSIs, so LR subspace estimation (LRSE) necessary when using RPCA, which complicated time-consuming. To solve this problem, letter proposes novel LR-based for called two-branch network combined...