Object detection has made tremendous progress in natural images over the last decade. However, results are hardly satisfactory when image object algorithm is directly applied to satellite images. This due intrinsic differences scale and orientation of objects generated by bird’s-eye perspective photographs. Moreover, background complex area small; as a result, small tend be missing challenge feature extraction. Dense overlap occlusion also affects performance. Although self-attention...

Hyperspectral images are widely used for classification due to its rich spectral information along with spatial information. To process the high dimensionality and nonlinearity of hyperspectral images, deep learning methods based on convolutional neural network (CNN) in applications. However, most CNN structures stacked vertically addition using a onefold size kernels or pooling layers, which cannot fully mine multiscale images. When such networks meet practical challenge limited labeled...

A compact, static hyperspectral imaging linear polarimeter (HILP) based on a Savart interferometer (SI) is conceptually described. It improves the existing SI by replacing front polarizer with two Wollaston prisms, and can simultaneously acquire four interferograms corresponding to linearly polarized lights single CCD. The spectral dependence of Stokes parameters be recovered Fourier transformation. Since there no rotating or moving parts, system relatively robust. interference model HILP...

Single-snapshot full-Stokes imaging polarimetry is a powerful tool for the acquisition of spatial polarization information in real time. According to general linear model polarimeter, recover full Stokes parameters at least four polarimetric intensities should be measured. In this paper, types single-snapshot division-of-aperture polarimeter with subapertures are presented and compared, maximum resolution each image on single area-array detector. By using error propagation theories different...

A high throughput static channeled interference imaging spectropolarimeter (CIISP) over 480-960nm spectral range is presented. The CIISP system includes two birefringent retarders and a Savart interferometer employing tempo-spatially mixed modulated mode with no internal moving parts, offers robust optical to resist the instrument noise. layout operation of sensor are presented in addition radiometric, improved polarimetric calibration techniques used system. performance verified through...

Change detection (CD) aims to identify differences in scenes observed at different times. Hyperspectral image (HSI) is preferred for the understanding of land surface changes, since it can provide essential and unique features CD. However, due high-dimensionality limited data, HSI-CD task challenged. While model-driven CD methods are hard achieve high accuracy weak performance fine data-driven be generalized data sets. The state-of-art method combine a single with convolutional neural...

For hyperspectral image change detection (HSI-CD) task, unsupervised learning methods based on pseudo-labels obtained from pre-classification method are promising due to its simplicity and generality. However, the inevitable noise lack of diversity in would degrade performance network model, resulting insufficient mining fine information HSIs. To address this issue, we propose a progressive pseudo-label (PPL) framework for HSI-CD. Differing state-of-the-art training designed networks with...

Hyperspectral target detection (HTD) aims to identifying targets within a hyperspectral image (HSI) based on provided spectra. In the current HTD field, representation-based detectors have attracted much attention. However, there are two prominent challenges that particularly noteworthy. First, background class encompasses diverse land covers, making its accurate representation challenging. Second, ability can be significantly influenced by abnormalities and noise in HSI. To tackle these...

Abstract In this paper, the design and experimental demonstration of a snapshot linear-Stokes imaging spectropolarimeter (SLSIS) is presented. The SLSIS, which based on division-of-focal-plane polarimetry with four parallel linear polarization channels integral field spectroscopy numerous slit dispersive paths, has no moving parts provides video-rate Stokes-vector hyperspectral datacubes. It does not need any scanning in spectral, spatial or dimension offers significant advantages rapid...

In this Letter, a novel snapshot spectral imaging technique, optically replicating and remapping spectrometer, is presented. It based on the combination of shifting subimages by specially designed lenslet array (LA) filtering focal plane continuous variable filter (CVF). The 3D datacube recovered just using simple image process. use LA CVF makes system compact low in cost. A handheld proof-of-principle prototype has been built demonstrated; it covers wavelength range 380-860 nm with 80...

A snapshot full-Stokes division-of-aperture imaging polarimeter using a Wollaston prism array (WPA) is theoretically described and experimentally demonstrated. Two-dimensional spatial distributions of six polarization eigenstates, linear (0°, 90°, 45°, 135°), left right circular states, are identified separated by the WPA simultaneously projected onto portions single focal-plane lens array. The conditions measurement matrix formed modulation channels naturally superior for immunity to...

An achromatic snapshot full-Stokes imaging polarimeter (ASSIP) that enables the acquisition of 2D-spatial full Stokes parameters from a single exposure is presented. It based on division-of-aperture polarimetry using an array four-quadrant elliptical analyzers as polarization state analyzer (PSA). The optimization PSA addressed for achieving immunity Gaussian and Poisson noises. extended eigenvalue calibration method (ECM) proposed to calibrate system, which considers imperfectness retarder...

A polarization imaging spectrometer based on a modified Savart polariscope with moving wedge prism is presented. The principle of the instrument described, and optical path difference as function prism's displacement calculated analyzed. It employs common-path configuration not sensitive to nonuniform variation speed environmental vibrations. In comparison polariscope, this framing rather than pushbrooming device. Only transmission birefringent materials detector sensitivity limit available...

A static polarization-difference imaging spectrometer is conceptually described and demonstrated through experiment. It consists of a Wollaston prism, Savart polariscope, linear analyzer, CCD camera. This design improves the existing system by eliminating its moving parts obtaining spectral variation polarization state, making more compact robust. After simultaneously acquiring two sequential interference images corresponding to orthogonal states, hyperspectral states can be reconstructed,...

In this Letter, the focal spot areas of an azimuthally polarized beam modulated with a vortex-0-2π-phase plate or π-phase-step are numerically found to be smaller than radially for three pupil functions uniform, Gaussian, and Bessel-Gauss profiles. Several types multizone phase plates theoretically designed simulated generating tight multifocal arrays from beams what we believe is first time. The positions polarization states can controlled simply by varying pattern plates. produced array...

Small target detection is the primary technology for infrared search and tracking (IRST) systems plays a vital role in practical applications. Existing algorithms have following challenges: 1) insufficient local nonlocal feature extraction 2) imbalance between accuracy real-time performance. In this study, novel model fast based on interpatch correlation enhancement (IPCE) joint visual saliency prior proposed to overcome such issues. Regarding dimension, improved tensor nuclear norm used...

Compressive full-Stokes spectropolarimetric imaging (SPI), integrating passive polarization modulator (PM) into general spectrometer, is powerful enough to capture high-dimensional information via incomplete measurement; a reconstruction algorithm needed recover 3D data cube (x, y, and λ) for each Stokes parameter. However, existing PMs usually consist of complex elements enslave accurate calibration, current algorithms suffer from poor quality are subject noise perturbation. In this work,...

For a full-Stokes polarimeter (FSP), generally there are two types of noise, signal-dependent Poisson shot noise and signal-independent additive Gaussian which will degrade the signal-to-noise ratio on measured Stokes parameters. The relation between immunity to condition measurement matrix has been widely studied in recent literature. In this paper, we present new merit function use it achieve optimal configurations with both noise. numerical results show that, for FSP consisting variable...

Based on empirical mode decomposition (EMD), the background removal and de-noising procedures of data taken by polarization interference imaging interferometer (PIIS) are implemented. Through numerical simulation, it is discovered that processing methods effective. The assumption noise mostly exists in first intrinsic function verified, parameters EMD thresholding determined. In comparison, wavelet windowed Fourier transform based introduced. de-noised results evaluated SNR, spectral...

In this paper, the design, optimization and analysis of broadband full-Stokes polarimeters with immunity to both Poisson Gaussian noise are presented. Different from commonly-used metrics such as, condition number (CN), equally weighted variance (EWV), or polarimetric modulation efficiency (PME) for noise, optimally balanced (BCPN) is, first time, proposed used as a metric polarimeters. The numerical results show that optimized BCPN have noise. achromatic similar properties over waveband, in...