Due to the advantages of deep learning, in this paper, a regularized feature extraction (FE) method is presented for hyperspectral image (HSI) classification using convolutional neural network (CNN). The proposed approach employs several and pooling layers extract features from HSIs, which are nonlinear, discriminant, invariant. These useful target detection. Furthermore, order address common issue imbalance between high dimensionality limited availability training samples HSI, few...

Classification is one of the most popular topics in hyperspectral remote sensing. In last two decades, a huge number methods were proposed to deal with data classification problem. However, them do not hierarchically extract deep features. this paper, concept learning introduced into for first time. First, we verify eligibility stacked autoencoders by following classical spectral information-based classification. Second, new way classifying spatial-dominated information proposed. We then...

Hyperspectral data classification is a hot topic in remote sensing community. In recent years, significant effort has been focused on this issue. However, most of the methods extract features original shallow manner. paper, we introduce deep learning approach into hyperspectral image classification. A new feature extraction (FE) and framework are proposed for analysis based belief network (DBN). First, verify eligibility restricted Boltzmann machine (RBM) DBN by following spectral...

A generative adversarial network (GAN) usually contains a and discriminative in competition with each other. The GAN has shown its capability variety of applications. In this paper, the usefulness effectiveness for classification hyperspectral images (HSIs) are explored first time. proposed GAN, convolutional neural (CNN) is designed to discriminate inputs another CNN used generate so-called fake inputs. aforementioned CNNs trained together: tries that as real possible, classify This kind...

Hyperspectral image (HSI) classification has become a hot topic in the field of remote sensing. In general, complex characteristics hyperspectral data make accurate such challenging for traditional machine learning methods. addition, imaging often deals with an inherently nonlinear relation between captured spectral information and corresponding materials. recent years, deep been recognized as powerful feature-extraction tool to effectively address problems widely used number processing...

In recent years, airborne and spaceborne hyperspectral imaging systems have advanced in terms of spectral spatial resolution, which makes the data sets they produce a valuable source for land cover classification. The availability with fine resolution has revolutionized image (HSI) classification techniques by taking advantage both information single framework.

Recently, a great many deep convolutional neural network (CNN)-based methods have been proposed for hyperspectral image (HSI) classification. Although the CNN-based advantages of spatial feature extraction, they are difficult to handle sequential data with and CNNs not good at modeling long-range dependencies. However, spectra HSI kind data, usually contains hundreds bands. Therefore, it is processing well. On other hand, Transformer model, which based on an attention mechanism, has proved...

Recently, the capability of deep learning-based approaches, especially convolutional neural networks (CNNs), has been investigated for hyperspectral remote sensing feature extraction (FE) and classification. Due to large number learnable parameters in filters, lots training samples are needed CNNs avoid overfitting problem. On other hand, Gabor filtering can effectively extract spatial information including edges textures, which may reduce FE burden CNNs. In this letter, order make most CNN...

The multisensory fusion of remote sensing data has obtained a great attention in recent years. In this letter, we propose new feature framework based on deep neural networks (DNNs). proposed employs convolutional (CNNs) to effectively extract features multi-/hyperspectral and light detection ranging data. Then, fully connected DNN is designed fuse the heterogeneous by previous CNNs. Through aforementioned networks, one can discriminant invariant data, which are useful for further processing....

One of the challenges in hyperspectral image (HSI) classification is that there are limited labeled samples to train a classifier for very high-dimensional data. In practical applications, we often encounter an HSI domain (called target domain) with few data, while another source may have enough Classes between two domains not be same. This article attempts use class data help classify classes, including same and new unseen classes. To address this paradigm, meta-learning paradigm few-shot...

Hyperspectral image (HSI) classification is a core task in the remote sensing community, and recently, deep learning-based methods have shown their capability of accurate HSIs. Among methods, convolutional neural networks (CNNs) been widely used for HSI classification. In order to obtain good performance, substantial efforts are required design proper learning architecture. Furthermore, manually designed architecture may not fit specific data set very well. this paper, idea automatic CNN...

Deep convolutional neural networks (CNNs) have shown their outstanding performance in the hyperspectral image (HSI) classification. The success of CNN-based HSI classification relies on availability sufficient training samples. However, collection samples is expensive and time consuming. Besides, there are many pretrained models large-scale data sets, which extract general discriminative features. proper reusage low-level midlevel representations will significantly improve accuracy. ImageNet...

Object detection in remote-sensing images (RSIs) is always a vibrant research topic the community. Recently, deep-convolutional-neural-network (CNN)-based methods, including region-CNN-based and You-Only-Look-Once-based have become de-facto standard for RSI object detection. CNNs are good at local feature extraction but they limitations capturing global features. However, attention-based transformer can obtain relationships of long distance. Therefore, Transformer Remote-Sensing (TRD)...

Accurate classification of remote sensing (RS) images is perennial topic interest in the RS community. Recently, transfer learning, especially for fine-tuning pre-trained convolutional neural networks (CNNs), has been proposed as a feasible strategy scene classification. However, because target domain (i.e., images) and source (e.g., ImageNet) are quite different, simply using model on an ImageNet dataset presents some difficulties. The models need to be properly adjusted build better...

Existing oriented object detection in aerial images has progressed a lot recent years and achieved favorable success. However, high-precision remains challenging task. Some works have adopted the classification-based method to predict angle order address boundary problem angle. we found that these often neglect sensitivity of objects with different aspect ratios At same time, it is worth exploring suitable way improve emerging transformer-based approaches adapt them detection. In this paper,...

Recently, transformer has gradually attracted interest for its excellence in modeling the long-range dependencies of spatial-spectral features HSI. However, problem quadratic computational complexity due to self-attention mechanism, which is heavier than other models and thus limited adoption HSI processing. Fortunately, recently emerging state space model-based Mamba shows great efficiency while achieving power transformers. Therefore, this paper, we first proposed spectral-spatial...

In existing convolutional neural networks (CNNs), both convolution and pooling are locally performed for image regions separately, no contextual dependencies between different have been taken into consideration. Such represent useful spatial structure information in images. Whereas recurrent (RNNs) designed learning among sequential data by using the (feedback) connections. this work, we propose network (C-RNN), which learns to enhance discriminative power of representation. The C-RNN is...

In this letter, a self-improving convolutional neural network (CNN) based method is proposed for the classification of hyperspectral data. This approach solves so-called curse dimensionality and lack available training samples by iteratively selecting most informative bands suitable designed via fractional order Darwinian particle swarm optimization. The selected are then fed to system produce final map. Experimental results have been conducted with two well-known data sets: Indian Pines...

Deep learning models, especially deep convolutional neural networks (CNNs), have been intensively investigated for hyperspectral image (HSI) classification due to their powerful feature extraction ability. In the same manner, ensemble-based systems demonstrated high potential effectively perform supervised classification. order boost performance of learning-based HSI classification, idea ensemble framework is proposed here, which loosely based on integration model and random subspace-based...

Deep convolutional neural networks (CNNs) have shown their great success on image classification. CNNs mainly consist of and pooling layers, both which are performed local areas without considering the dependence among different regions. However, such is very important for generating explicit representation. In contrast, recurrent (RNNs) well known ability encoding contextual information in sequential data, they only require a limited number network parameters. Thus, we proposed hierarchical...

Capsule networks can be considered to the next era of deep learning and have recently shown their advantages in supervised classification. Instead using scalar values represent features, capsule use vectors which enriches feature presentation capability. This paper introduces a network for hyperspectral image (HSI) classification improve performance conventional convolutional neural (CNNs). Furthermore, modification named Conv-Capsule is proposed. full connections, local connections shared...

In recent years, deep convolutional neural networks (CNNs) have been widely used for hyperspectral image (HSI) classification. The powerful feature extraction capability and high classification performance of CNN highly depend on sufficient training samples. Unfortunately, it is not a common situation because collecting samples time-consuming expensive. this letter, in order to make the most with limited samples, dual-path siamese (Dual-SCNN) proposed HSI Specifically, framework combination...

Jamming is a big threat to radar system survival and anti-jamming part of the solution. The classification jamming signal first step toward anti-jamming. Recently, as an important deep learning, convolutional neural network (CNN) based methods have shown their capability in discriminant feature extraction accurate classification. In this study, order harness powerfulness CNN are proposed classify acting on pulse compression radar. Specifically, 1D-CNN designed for under condition sufficient...