Few-Shot Action Recognition (FSAR) is a challenging task that requires recognizing novel action categories with a few labeled videos. Recent works typically apply semantically coarse category names as auxiliary contexts to guide the learning of discriminative visual features. However, such context provided by the action names is too limited to provide sufficient background knowledge for capturing novel spatial and temporal concepts in actions. In this paper, we propose DiST, an innovative Decomposition-incorporation framework for FSAR that makes use of decoupled Spatial and Temporal knowledge provided by large language models to learn expressive multi-granularity prototypes. In the decomposition stage, we decouple vanilla action names into diverse spatio-temporal attribute descriptions (action-related knowledge). Such commonsense knowledge complements semantic contexts from spatial and temporal perspectives. In the incorporation stage, we propose Spatial/Temporal Knowledge Compensators (SKC/TKC) to discover discriminative object-level and frame-level prototypes, respectively. In SKC, object-level prototypes adaptively aggregate important patch tokens under the guidance of spatial knowledge. Moreover, in TKC, frame-level prototypes utilize temporal attributes to assist in inter-frame temporal relation modeling. These learned prototypes thus provide transparency in capturing fine-grained spatial details and diverse temporal patterns. Experimental results show DiST achieves state-of-the-art results on five standard FSAR datasets.

Space-borne multi-angle polarization remote sensing is considered to be one of the most important tools to obtain global aerosol parameters in assessment of climate change. Accurate calibration is a prerequisite for quantitative polarization remote sensing. However, most research focuses on the polarization calibration methods and the monitoring of the calibration coefficient stability. Few studies investigate the polarization calibration verification of the subsequent new satellite sensors. The Directional Polarimetric Camera (DPC) onboard the Chinese Terrestrial Ecosystem Carbon Inventory Satellite (abbreviated as TECIS, with the Chinese name ”Gou Mang”) is a brand-new polarization sensor. To evaluate the polarization calibration of this sensor, we propose a verification scheme for polarization calibration, which can verify both the degree of polarization (DoP) and polarized reflectance. The accuracy of DoP is verified by using sunglint on the ocean. When detecting the sunglint, constraints such as observational geometry, cloud identification, and wind speed are introduced, and the polarized reflectance is atmospherically corrected according to the marine aerosol model and aerosol optical depth from MODIS. The accuracy of polarized reflectance at the Top of Atmosphere (TOA) is verified based on AERONET inversion products and the Bidirectional Polarization Distribution Functions (BPDF). Experiments show that the accuracies of DoP of the three polarization channels (490, 670, 865 nm) of DPC are 5.57%, 2.07%, and 1.97% respectively, and the average accuracy of the multi-angle polarized reflectance at the TOA of the 865nm channel is 0.209%.

Few-Shot Action Recognition (FSAR) aims to recognize novel class action with limited annotated training data from the same class. Most FSAR methods subconsciously follow the few-shot image classification solutions by solely focusing on appearance-level matching between support and query videos, such as part-level matching, frame-level matching, and segment-level matching. However, these methods, almost always, have two main limitations: 1) generally ignore the relationship among these part-, frame- and segment-level features and 2) may mismatch the same class actions under fast-term and slow-term dynamics. To this end, we present a novel Hierarchical Motion-enhanced Matching (HM${^{2}}$) framework to hierarchically learn the relation-aware multi-modal features, and jointly promote the multi-modal matching, including appearance-level matching on segments, frames, and parts, as well as the motion-level matching on dynamics. Specifically, we first propose a new Hierarchical Tokenizer (HT) to learn multi-modal features, namely utilizing a hierarchical Transformer to learn appearance-level features, along with a Slow-Fast Aware Motion (SFAM) strategy to learn motion-level features covering fast- and slow-term dynamics. Next, we propose a new Relation-aware Matcher (RM) to match the multi-modal features, by leveraging a Hierarchical Relational Graph Convolutional Network (H-RGCN) to capture the relationship among these appearance-level features. Further, a Dual Sample-to-Class Matching (DSCM) strategy is proposed to measure the bidirectional similarities among appearance- and motion-modal features by sample-to-class matching and class-to-sample matching. Extensive experiments on four golden FSAR datasets demonstrate significant performance improvements of HM${^{2}}$compared with the state-of-the-art methods.

Recent few-shot action recognition (FSAR) methods typically perform semantic matching on learned discriminative features to achieve promising performance. However, most FSAR methods focus on single-scale (e.g., frame-level, segment-level,etc.) feature alignment, which ignores that human actions with the same semantic may appear at different velocities. To this end, we develop a novel Multi-Velocity Progressive-alignment (MVP-Shot) framework to progressively learn and align semantic-related action features at multi-velocity levels. Concretely, a Multi-Velocity Feature Alignment (MVFA) module is designed to measure the similarity between features from support and query videos with different velocity scales and then merge all similarity scores in a residual fashion. To avoid the multiple velocity features deviating from the underlying motion semantic, our proposed Progressive Semantic-Tailored Interaction (PSTI) module injects velocity-tailored text information into the video feature via feature interaction on channel and temporal domains at different velocities. The above two modules compensate for each other to make more accurate query sample predictions under the few-shot settings. Experimental results show our method outperforms current state-of-the-art methods on multiple standard few-shot benchmarks (i.e., HMDB51, UCF101, Kinetics, SSv2-full, and SSv2-small).

The panchromatic and multispectral (PMS) sensor is a high spatial resolution sensor aboard the GF-1 satellite launched on April 26, 2013. This paper focuses on the cross-calibration of the PMS sensor using Terra/Moderate-Resolution Imaging Spectroradiometer (MODIS) and Landsat 8/Operational Land Imager (OLI). Two matched-image adjustment factors (MIAFs) are used in the cross-calibration which are the radiance MIAF and reflectance MIAF. Two test sites are chosen as the regions of interest. One is the Dunhuang test site, which has been used for the vicarious calibration of Chinese satellites since later 1990s. The other is the Golmud test site, which is a new site with no ground measured data available. The results show that both the Dunhuang and Golmud test sites can be used for cross-calibration. This paper reveals that the cross-calibration of the PMS sensor using OLI is better than using MODIS, as the calibration coefficient difference between the two test sites with OLI is smaller than that with MODIS. The uncertainty analysis results show that the uncertainty of cross-calibration using OLI is 5%-7% when the ground data are not available.

This paper describes a methodology that uses Hyperion imagery as reference data to calibrate the Chinese Hyperspectral Imager (HSI) onboard the HJ-1A satellite. Two test sites near Dunhuang in Gansu and in Inner Mongolia were used for the cross-calibration. To account for the uncertainties in the top of atmosphere (TOA) reflectance due to the bidirectional reflectance distribution function (BRDF) and relative spectral response (RSR) differences between the two sensors, a model is adopted to transfer the Hyperion TOA reflectance to the effective HSI TOA reflectance. The influence of BRDF is analyzed, and two BRDF correction approaches are applied to the two test sites, respectively. For the Dunhuang test site, the ground synchronously measured reflectance in different solar zenith angles is used for BRDF correction. For the Inner Mongolia test site, a kernel-driven model is applied. The influence of RSR mismatch is computed using a spectral profile adjustment factor (SPAF), which takes into account the spectral profile of the target and the RSR of each sensor. The SPAF is calculated according to the TOA reflectance simulated using moderate resolution atmospheric transmission. One-point calibration and multipoint calibration coefficients are computed, respectively. Ground reflectance data measured in June 2010 at the Inner Mongolia test site were used to validate the cross-calibration coefficients based on the Hyperion image. The results support the proposal that the cross-calibration method between two hyperspectral sensors is effective, and the use of multipoint calibration coefficient with nonzero offset has great potential for hyperspectral sensor calibration.

In this paper, a model for light interaction has been developed to compute bidirectional reflectance from realistic 3D canopies approximated by an arbitrary configuration of plants. It can well simulate multi-spectral reflectance of semi-arid nature grassland. There are two important parts of the simulation model. The first part is the generation of the 3D realistic grassland scene. In this model, the Clumped Architecture Model of Plants (CLAMP) is used to. The second part is the determining the visibility and brightness of grass canopy scene using Geometric Optics Model. The simulating model is validated by comparing the simulation result with the HJ satellite data at synchronous time. As a result, it can describe directional reflectance properties of semi-grassland canopies in terms of canopy architecture parameters and optical scattering properties of discrete phytoelements.

FY1C is a polar meteorological satellite of China, which had been worked on orbit about 5 years. In this paper, time series calibration method based on pseudo-invariant target site is applied to monitor the variance of FY1C instrument. Dunhuang test site is chose as the pseudo-invariant site and the FY1C images over this site are processed with some standard. Then the time series calibration result of FY1C seven channels at visible and near-infrared range has been calculated. In order to validate the result, apply the time series calibration coefficients to recalibrate the images of Wuwei test site from 1999 to 2003. The validation result shows that the time series calibration coefficients are efficient and can monitor the radiance status of FY1C instrument.

CBERS02B with three payloads onboard was successfully launched on September 19, 2007 in order to ensure the continuity of CBERS series and CCD is one of three payloads. Calibration of CCD is a precursor for its quantitative application because there isn't onboard calibrator for CCD. A comprehensive vicarious calibration and validation campaign of CCD was performed at Gongger test site on October 12, 2007. The reflectance-based calibration method was used in this campaign with the ground measurements of the surface reflectance and atmospheric characteristics. Then 6S, a radiative transfer code, was used to compute the top-of-atmosphere(TOA) radiance at the sensor. Calibration result was obtained for CCD showing that some change brought to the CCD after launch, especially band 1 and band 2. The in-situ field measurement at the Dunhuang test site was collected validating that the calibration result was good expect for band 4.

HJ-1A satellite has been lunched in September, 2008. It is calibration and validation that the fundamental of quantitative utilization of HJ-1A IRS imagery. HJ-1A has only one channel in thermal infrared band, compared to MODIS sensor, which has two channels accordingly. The key process of cross-calibration is band match, so this paper uses the MODIS SST product retrieval algorithm as reference for the difference of HJ-1A and MODIS thermal infrared channels characters. TIGR database were used as input parameters into radiative transfer mode Modtran4.0 to obtain band match coefficients by regression analysis. Number 711~1064 datum in TIGR database representing mid-latitude winter were chose according to the selected image's date. Research demonstrates that cross-calibration method is effective to HJ-1A thermal infrared channel 4.

A good calibration of satellite is necessary to derive reliable quantitative measurements of the surface parameters or to compare data obtain from different sensors. DCSRS (Demonstration Center for Spaceborne Remote Sensing of China National Space Administration) went to inner-Mongolia Plateau to seek fairly uniform reflectance sites as a part of Beijing multi functional test site network in May and October 2006, and four quite flat and homogenous sites were selected as potential test sites. These four sites have many good calibration site characteristics: they are large and flat; the rain is little and the evaporation is much larger than precipitation, so the water vapor content is little in atmosphere; the elevation is about 1100m and the weather is sunny in most time. In this study, more than 200 MODIS level 1 images of these four sites were obtained and the average reflectance and relative mean squared deviation of each image were calculated. In the end, the variation of reflectance with solar zenith, month and season were analyzed, and the result was consistent with in-situ investigation.

In this paper, a synthetic strategy has been employed to model 3D canopy's radiation transfer in the whole optical spectral domains. 3D plant architecture model (the Clumped Architecture Model of Plants: CLAMP) (1) is used to generate the realistic vegetation scene. In the visible and NIR region, the canopy BRDF was decomposed into three parts: single scattering contribution from leaves, single scattering contribution from the soil, and multiple scattering part of the canopy. The single scattering contributions come from illuminated leaves and soil components which are computed by the reverse ray-tracing procedure (2) with their corresponding reflectance. The multiple scattering contribution is approximated by the four-stream theory. As a result, the modeling of VNIR region is more efficient and fairly accurately describes the anisotropically scattering features of vegetation. In the TIR region, the directional brightness temperature of canopy is calculated as the linear combination of four component's (illuminated leaves, illuminated ground, shadowed leaves, and shadowed ground) brightness temperature multiplied by its fractional cover computed by the reverse ray-tracing procedure. Initial modeling results show typical features of vegetation's anisotropic scattering and directional temperature distributions, for example, hot spot, bowl shape and reach a good agreement with theoretical results in those three domains. This strategy shows potential of exploring the impact of canopy structure on the radiometric response measured by remote sensors.