In this article, we present a new pansharpening method, zero-reference generative adversarial network (ZeRGAN), which fuses low spatial resolution multispectral (LR MS) and high panchromatic (PAN) images. the proposed indicates that it does not require paired reduced-scale images or unpaired full-scale for training. To obtain accurate fusion results, establish an game between set of multiscale generators their corresponding discriminators. Through generators, fused MS (HR are progressively...

The integration of spatial and spectral information is beneficial to the improvement change detection (CD) performance. However, existing methods cannot efficiently suppress influences differences (SDs) in unchanged areas. To address these issues, this article, we propose a content-guided spatial–spectral network (CSI-Net) for fusion global details SD information. Specifically, proposed CSI-Net composed reasoning (SR) module, an (CGI) module. In SR learned by cascaded graph convolution (GC)...

Recently, deep neural network (DNN)-based methods have achieved good results in terms of the fusion low spatial resolution hyperspectral (LR HS) and high multispectral (HR MS) images. However, spectral band correlation (SBC) nonlocal similarity (SNS) (HS) images are not sufficiently exploited by them. To model two priors efficiently, we propose a spectral-spatial dual graph unfolding (SDGU-Net), which is derived from optimization regularized restoration models. Specifically, introduce graphs...

Pose distillation is widely adopted to reduce model size in human pose estimation. However, existing methods primarily emphasize the transfer of teacher knowledge while often neglecting performance degradation resulted from curse capacity gap between and student. To address this issue, we propose AgentPose, a novel method that integrates feature agent distribution features progressively aligns student with feature, effectively overcoming enhancing ability transfer. Our comprehensive...

Pan-sharpening methods based on deep neural network (DNN) have produced state-of-the-art fusion performance. However, DNN-based mainly focus the modeling of local properties in low spatial resolution multispectral (LR MS) and panchromatic (PAN) images by convolution networks. The global dependencies are ignored. To capture concurrently, we propose a multiscale spatial–spectral interaction transformer (MSIT) for pan-sharpening. Specifically, construct sub-networks containing...

Recently, pan-sharpening methods based on deep learning (DL) have achieved state-of-the-art results. However, current existing DL-based need to be trained repetitively for different satellite sensors obtain satisfactory fusion performance and therefore require a large number of training images each satellite. To deal with these issues, in this paper we propose unified two-stage spatial spectral network (UTSN) pan-sharpening. A branch networks is constructed satellite, which the enhancement...

Currently, convolution neural networks (CNNs) and transformers have been the dominant paradigms for change detection (CD) thanks to their powerful local global feature extraction capabilities. However, with improvement of resolution, spatial, spectral, temporal relationships among objects in remote sensing images are becoming more complicated, cannot be modeled efficiently by existing methods. To capture high-order complex images, we propose a multiview hypergraph fusion network (MVHFNet)...

In this article, we proposed a novel image fusion method based on multiscale convolution sparse decomposition (MCSD). A unified framework MCSD is first utilized to decompose panchromatic (PAN) and the spatial component of upsampled low resolution multispectral (LR MS) images, which can produce corresponding frequencies feature maps. By combining with analysis, efficiently approximate spectral information in images. Next, binary map generated from gradient integrate LR MS PAN For maps, gain...

The majority of existing hyperspectral (HS) image denoising methods exploit local similarity in HS images by rearranging them into the matrix or vector forms. As typical 3-D data, inherent spatial and spectral properties should be simultaneously explored for denoising. Therefore, a geometrical kernel (3DGK) is developed this article to describe structure. proposed method assumes that pixel can represented other pixels within block efficiently owing with adjacent positions. Then, modeled...

In this article, we propose a new pan-sharpening method that disentangles low spatial resolution multispectral (LRMS) and panchromatic (PAN) images in terms of sensor-specific features common features. These are obtained by defining mutual information (MI)-based transformers designed to achieve disentangled learning. the proposed method, LRMS PAN cross-reconstructed cross-coupled facilitate disentanglement To ensure compatibility among features, self-reconstructions imposed on them, source...

Microelectronics circuit analysis and design is one of the most important courses for all undergraduate majors in School Electrical Engineering.However, traditional lecture teaching method suffers from disadvantage low efficiency, large amount time consuming passive learning habits.This paper discusses use project based approach to improve quality teaching/learning analog design.The proposed methodology has been applied several electrical engineering courses.The allows students better...

Existing deep neural network (DNN)-based image fusion methods seldom consider low-rank priors for the decomposition of source images, which cannot efficiently model base and detail components in images. To exploit better, we propose a rank- <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</i> (DRDec-Net) according to Specifically, is first established by imposing on component Then, based model, construct DRDec-Net, composed (LRD) modules,...

Ensuring safety on the road is crucial, and detecting driving distractions plays a vital role in achieving this goal. Accurate identification of distracted behaviors facilitates prompt intervention, thereby contributing to reduction accidents. We introduce an advanced fully convolutional one-stage (FCOS) object detection algorithm tailored for distraction that leverages knowledge distillation framework. Our proposed methodology enhances conventional FCOS through integration selective kernel...

Depth provided by binocular vision systems is extra informationfor stereo visual tracking, which has less distance limitation andcan work well in outdoor environment compared to red-green-blue-depth (RGB-D) cameras.

Traditionally the singularity point in extreme scenarios such as weak traveling waves may not be detected accurately a result of their small amplitudes and naturally-existing attenuation. In this paper, novel method based on curve fitting theory for detection (weak waves) is proposed, which considers both characteristic influence distributed capacitance at bus bar. Mathematical equation rising edge initial wave deduced paper. Furthermore, indirect mathematical Maclaurin expansion derived to...

CCD image usually has a huge amount of data with high-resolution, therefore it is difficult to capture and browse the real-timely in real project. In traditional transportation systems, we first store into storage medium, then read out for later, but this method not good diagnosis debugging real-timely. For eliminating disadvantages, advance new platform system, based on serial SRIO interface, captured can be transported immediately at line speed 3.125Gb/s displayed rapidly. Since its...