A simple cloud masking approach using NOAA AVHRR daytime data for tropical areas

A simple method for cloud detection for AVHRR daytime data is presented and checked for equatorial/tropical areas based on a study area in northeastern Brazil. Five different cloud masking techniques based on visible and infrared spectral information for cloud detection are calibrated. The significant differences between the equatorial threshold obtained in this work and the midlatitude thresholds given by Saunders and Kriebel and by Thiermann and Ruprecht are compared and discussed. Results from the cloud masking algorithm are presented and comments are made about the problems related to the automatic cloud detection algorithm presented in this study.     

Absolute calibration of AVHRR visible and near-infrared channels using ocean and cloud views

Methods for absolute calibration of visible and near-infrared sensors using ocean and cloud views have been developed and applied to channels 1 (red) and 2 (near-infrared) of the Advanced Very High Resolution Radiometer (AVHRR) for the NOAA-7, -9 and -11 satellites. The approach includes two steps. First step is intercalibration between channels 1 and 2 using high altitude (12 km and above) bright clouds as ‘ white’ targets. This cloud intercalibration is compared with intercalibration using ocean glint. The second step is an absolute calibration of channel 1 employing ocean off-nadir view (40-70° ) in channels 1 and 2 and correction for the aerosol effect. In this process the satellite measurements in channel 2, corrected for water vapour absorption are used to correct channel 1 for aerosol effect. The net signal in channel 1 composed from the predictable Rayleigh scattering component is used to calibrate this channel. The result is an absolute calibration of the two AVHRR channels. NOAA-9 channels I and 2 show a degradation rate of 8-8 per cent and 6 per cent, respectively, during 1985-1988 and no further degradation during 1988-1989 period. NOAA-II shows no degradation during the 1989 mid 1991 period. This trend is similar to the calibration trend obtained using desert site observations, the absolute calibration found in this work for both sensors is lower by 17 to 20 per cent ( suggesting higher degradation) from the absolute calibration of Abel et al. ( 1993 Journal of Atmospheric and Ocean Technology,10, 493-508 that used aircraft measurements. Furthermore we show that application of the calibration of Abel et al. or the present one for remote sensing of aerosol over Tasmania, Australia failed to predict correctly the aerosol optical thickness measured there. The only way to reconcile all these differences is by allowing for a shift of 17 nm towards longer wavelengths of the AVHRR channel 1 effective wavelength. We show that with this shift, we get an agreement between the two absolute calibration techniques ( ± 3 percent), and both of them do predict correctly the optical thickness in the two channels ( + 0.02) Recent work in preparation for publication (Vermote el al, 1995, in preparation indicates that this shift is due to an out of band transmission ( 6 per cent at 900nm) for AVHRR channel 1 previously unidentified.     

Generalized Bayesian cloud detection for satellite imagery. Part 1: Technique and validation for night-time imagery over land and sea

Numerical Weather Prediction (NWP) fields are used to assist the detection of cloud in satellite imagery. Simulated observations based on NWP are used within a framework based on Bayes' theorem to calculate a physically-based probability of each pixel with an imaged scene being clear or cloudy. Different thresholds can be set on the probabilities to create application-specific cloud-masks. Here, this is done over both land and ocean using night-time (infrared) imagery. We use a validation dataset of difficult cloud detection targets for the Spinning Enhanced Visible and Infrared Imager (SEVIRI) achieving true skill scores of 87% and 48% for ocean and land, respectively using the Bayesian technique, compared to 74% and 39%, respectively for the threshold-based techniques associated with the validation dataset.     

伪激光点云增强的道路场景三维目标检测

目的 针对激光雷达点云稀疏性导致小目标检测精度下降的问题,提出一种伪激光点云增强技术,利用图像与点云融合,对稀疏的小目标几何信息进行补充,提升道路场景下三维目标检测性能。方法 首先,使用深度估计网络获取双目图像的深度图,利用激光点云对深度图进行深度校正,减少深度估计误差;其次,采用语义分割的方法获取图像的前景区域,仅将前景区域对应的深度图映射到三维空间中生成伪激光点云,提升伪激光点云中前景点的数量占比;最后,根据不同的观测距离对伪激光点云进行不同线数的下采样,并与原始激光点云进行融合作为最终的输入点云数据。结果 在KITTI （Karlsruhe Institute of Technology and Toyota Technological Institute at Chicago）数据集上的实验结果表明,该方法能够提升多个最新网络框架的小目标检测精度,以典型网络SECOND（sparselyembedded convolutional detection）、MVX-Net （multimodal voxelnet for 3D object detection）、Voxel-RCNN为例,在困难等级下,三维目标检测精度分别获得8.65%、7.32%和6.29%的大幅提升。结论 该方法适用于所有以点云为输入的目标检测网络,并显著提升了多个目标检测网络在道路场景下的小目标检测性能。该方法具备有效性与通用性。     

On modelling and rendering ocean scenes

This paper is devoted to ocean images. First we propose a complete geometrical model accounting for refraction, diffraction, reflection, transmission and multiwave trains. Then we describe a specific algorithm for the rendering of coastal scenes. Copyright © 2000 John Wiley & Sons, Ltd.     

An automated scheme for the removal of cloud contamination from AVHRR radiances over western Europe

Abstract

Various cloud-detection schemes are applied to 1.1 km Advanced Very High Resolution Radiometer (AVHRR) day- and night-time data to determine an optimum automated scheme for deriving cloud-free radiances over both land and sea. A combination of the spatial coherence method at infrared wavelengths (11 μm) and dynamic visible threshold methods proved to be the most effective scheme for day-time use. Uniform thin cirrus (i.e. reflectance less than 15 per cent) was difficult to detect with all methods. Problems were also encountered over regions with a changing underlying surface type (e.g. coastal areas) where the automated scheme was not as effective as over uniform surfaces. At night a combination of the spatial coherence method and a scheme based on the differences in brightness temperatures between the three infrared channels at 37, II and 12 μm wavelength was successfully used. Results obtained by applying these algorithms to AVHRR data are presented and the different problems encountered with each algorithm are discussed.     

A daytime cloud detection algorithm for FY-3A/VIRR data

Cloud detection is essential for the retrieval of atmospheric and surface parameters and it directly impacts the quality of many satellite geophysical products used in weather, climate and environmental research. In this article, a daytime cloud detection algorithm based on multi-spectral thresholds is proposed to discriminate clouds from clear skies for the visible and infrared radiometer (VIRR), which is a key instrument on board the Chinese FengYun-3A (FY-3A) polar-orbiting meteorological satellite, launched 27 May 2008. The VIRR has ten bands in the wavelengths 0.43–12.5 μm and provides global observations of atmosphere, ocean and land in the visible and infrared regions of the spectrum. In this algorithm, the underlying surface is divided into five ecological types: snow/ice, desert, coastal, land and water, and seven spectral bands of the VIRR are used to indicate a level of confidence that the VIRR is observing clear skies. This algorithm also utilizes the 1.6 μm band and the difference between the 1.38 and 1.6 μm bands to respectively detect water cloud and high cloud. An example of cloud detection and a comparison with an official cloud masking product are given; the results show that this algorithm performs well and is better than the official algorithm in cloud detection.     

A cloud detection algorithm for AATSR data, optimized for daytime observations in Canada

To extract information about the Earth's surface from Earth Observation data, a key processing step is the separation of pixels representing clear-sky observations of land or water surfaces from observations substantially influenced by clouds. This paper presents an algorithm used for this purpose specifically for data from the AATSR sensor on ENVISAT. The algorithm is based on the structure of the SPARC cloud detection scheme developed at CCRS for AVHRR data, then modified, calibrated and validated for AATSR data. It uses a series of weighted tests to calculate per-pixel cloud presence probability, and also produces an estimate of cloud top height and a cloud shadow flag. Algorithm parameters have been optimized for daytime use in Canada, and evaluation shows good performance with a mean daytime kappa coefficient of 0.76 for the ‘cloud’/‘clear’ classification when compared to independent validation data. Performance is independent of season, and is a dramatic improvement over the existing AATSR L1B cloud flag for Canada. The algorithm will be used at CCRS for processing AATSR data, and will form the basis of similar processing for data from the SLSTR sensors on Sentinel-3.     

Mesoscale cloud pattern classification over ocean with a neural network using a new index of cloud variability

The purpose of this study is to determine the feasibility of a mesoscale (<300 km) cloud classification using infrared radiance data of satellite‐borne instruments. A new method is presented involving an index called the diversity index (DI), derived from a parameter commonly used to describe ecosystem variability. In this respect, we consider several classes of value ranges of standard deviation of the brightness temperature at 11 µm (σ_BT). In order to calculate DI for 128×128 km² grids, subframes of 8 km×8 km are superimposed to the satellite image, and then σ_BT is calculated for all 256 subframes and assigned to one of the classes. Each observed cloud pattern is associated with an index characterized by the frequency of σ_BT classes within the scene, representative of a cloud type. Classification of different clouds is obtained from Advanced Very High Resolution Radiometer (AVHRR)‐NOAA 16 data at 1 km resolution. Stratus, stratocumulus and cumulus are specifically recognized by this window analysis using a DI threshold. Then, a six‐class scheme is presented, with the standard deviation of the infrared brightness temperature of the entire cloud scene (σ_c) and DI as inputs of a neural network algorithm. This neural network classifier achieves an overall accuracy of 77.5% for a six‐class scheme, and 79.4% for a three‐class scheme, as verified against the analyses of nephanalists as verified against a cloud classification from Météo France. As an application of the proposed methodology, regional cloud variability over Pacific is examined using cloud patterns derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) carried aboard Earth Observing System (EOS) Terra polar orbiter platform, for February 2003 and 2004. The comparison shows regional change in monthly mean cloud types, associated with 2003 El Niño and 2004 neutral events. A significant increase in the occurrence of convective clouds (+15%) and a decrease in stratiform clouds (?10%) are observed between the two months.     

Cloud top phase determination from the fusion of signatures in daytime AVHRR imagery and HIRS data

An improved methodology for the retrieval of water vapour profiles from DMSP SSM /T-2 microwave sounder data has been demonstrated using cloud-top temperatures derived from NOAA AVHRR imagery as a constraint. However, the automated analysis of cloud-top temperature in AVHRR imagery is complicated by the presence of optically-thin cirrus clouds, since a component of the upwelling radiation from below passes unatttenuated to space. Therefore, cloud-top phase must first be determined to ensure the accurate specification of cloud-top temperature. In this paper, a new approach is presented for the specification of cloud-top phase in an operational environment. The methodology combines results from bi-spectral cloud tests for ice and water clouds in daytime AVHRR imagery with cloud-top pressure analyses based upon the CO2 slicing of HIRS data. The accuracy of the automated cloud-top phase analyses is measured quantitatively against manual analyses of the AVHRR imagery. It is concluded that the fusion of cloud signatures in AVHRR imagery and HIRS data improves the specification of cloud-top phase in the higher resolution imagery and reduces the ambiguity inherent in analyses based solely upon bi-spectral techniques.     

Generalized Bayesian cloud detection for satellite imagery. Part 2: Technique and validation for daytime imagery

Numerical Weather Prediction (NWP) fields are used to assist the detection of cloud in satellite imagery. Simulated observations based on NWP are used within a framework based on Bayes' theorem to calculate a physically-based probability of each pixel with an imaged scene being clear or cloudy. Different thresholds can be set on the probabilities to create application-specific cloud masks. Here, the technique is shown to be suitable for daytime applications over land and sea, using visible and near-infrared imagery, in addition to thermal infrared. We use a validation dataset of difficult cloud detection targets for the Spinning Enhanced Visible and Infrared Imager (SEVIRI) achieving true skill scores of 89% and 73% for ocean and land, respectively using the Bayesian technique, compared to 90% and 70%, respectively for the threshold-based techniques associated with the validation dataset.     

Rapid pedestrian detection in unseen scenes

In this paper, a rapid adaptive pedestrian detection method based on cascade classifier with ternary pattern is proposed. The proposed method achieves its goal by employing the following three new strategies: (1) A method for adjusting the key parameters of the trained cascade classifier dynamically for detecting pedestrians in unseen scenes using only a small amount of labeled data from the new scenes. (2) An efficient optimization method is proposed, based on the cross entropy method and a priori knowledge of the scenes, to solve the classifier parameter optimization problem. (3) In order to further speed up pedestrian detection in unseen scenes, each strong classifier in the cascade employs a ternary detection pattern. In our experiments, two significantly different datasets, AHHF and NICTA, were used as the training set and testing set, respectively. The experimental results showed that the proposed method can quickly adapt a previously trained detector for pedestrian detection in various scenes compared with other existing methods.     

Vision-based entrance detection in outdoor scenes

Doors are a significant object for the visually impaired and robots to enter and exit buildings. Although the accuracy of door detection is reported high in indoor scenes, it has become a difficult problem in outdoor scenes in computer vision. The reason may lie in the fact that such properties of a simple ordinary door such as handles, corners, and the gap between the door and the ground may not be visible due to the great variety of doors in outdoor environments. In this paper, we present a vision-based method for detecting building entrances in outdoor images. After extracting the lines and deleting the extra ones, regions between the vertical lines are specified and the features including height, width, location, color, texture and the number of lines inside the regions are obtained. Finally, some additional knowledge such as door existence at the bottom of the image, a reasonable height and width of a door, the difference between color and texture of the doors and those of the neighboring regions, and numerous lines on doors is used to decide on door detection. The method was tested on the eTRIMS dataset, door images from the ImageNet dataset, and our own dataset including doors of houses, apartments, and stores leading to acceptable results. The obtained results show that our approach outperforms comparable state-of-the-art approaches.     

Object-based cloud and cloud shadow detection in Landsat imagery

A new method called Fmask (Function of mask) for cloud and cloud shadow detection in Landsat imagery is provided. Landsat Top of Atmosphere (TOA) reflectance and Brightness Temperature (BT) are used as inputs. Fmask first uses rules based on cloud physical properties to separate Potential Cloud Pixels (PCPs) and clear-sky pixels. Next, a normalized temperature probability, spectral variability probability, and brightness probability are combined to produce a probability mask for clouds over land and water separately. Then, the PCPs and the cloud probability mask are used together to derive the potential cloud layer. The darkening effect of the cloud shadows in the Near Infrared (NIR) Band is used to generate a potential shadow layer by applying the flood-fill transformation. Subsequently, 3D cloud objects are determined via segmentation of the potential cloud layer and assumption of a constant temperature lapse rate within each cloud object. The view angle of the satellite sensor and the illuminating angle are used to predict possible cloud shadow locations and select the one that has the maximum similarity with the potential cloud shadow mask. If the scene has snow, a snow mask is also produced. For a globally distributed set of reference data, the average Fmask overall cloud accuracy is as high as 96.4%. The goal is development of a cloud and cloud shadow detection algorithm suitable for routine usage with Landsat images.     

Spatial and seasonal patterns of night-time lights in global ocean derived from VIIRS DNB images

ABSTRACT

The majority of the research using night-time data has focused on the terrestrial environment, while the light flooding our oceans is less studied. Meanwhile, given the rapid development of imaging technology at night, remotely sensed night-time light data can now provide a great opportunity to improve understanding of the spatiotemporal distribution of light over large areas. In this article, we used monthly cloud-free night-time imagery from the Suomi National Polar-Orbiting Partnership (Suomi NPP) Visible Infrared Imaging Radiometer Suite Day/Night Band, collected from 2014 to 2016, to explore the spatial distribution of night-time light in marine ecosystems. Morphological methods were used to extract light agglomeration areas. Using the Empirical Mode Decomposition method, we detected seasonal change patterns. Our results show that the distribution of light at night is clustered, and mainly concentrated in coastal and offshore waters, with about 70% of the total light found in 0.3% of the global marine waters. Flares from oil and gas well may not create a distinctive seasonal pattern, although fishing lights may show a seasonal pattern. The five largest agglomeration areas of light are centred in the eastern and southeast waters of Asia with little seasonal fluctuation. The cyclical light pattern of the entire marine system had a period of about 0.94 years, while varied from 0.5 to 1.1 years in the agglomeration areas. The proportions of seasonal energy for 49% of the top 100 agglomeration areas were below 10%, while the areas located in the waters near northern Japan, North Korea, eastern Indonesia, and eastern Argentina experienced large seasonal changes.     

杂乱场景下小物体抓取检测研究

目的 杂乱场景下的物体抓取姿态检测是智能机器人的一项基本技能。尽管六自由度抓取学习取得了进展,但先前的方法在采样和学习中忽略了物体尺寸差异,导致在小物体上抓取表现较差。方法 提出了一种物体掩码辅助采样方法,在所有物体上采样相同的点以平衡抓取分布,解决了采样点分布不均匀问题。此外,学习时采用多尺度学习策略,在物体部分点云上使用多尺度圆柱分组以提升局部几何表示能力,解决了由物体尺度差异导致的学习抓取操作参数困难问题。通过设计一个端到端的抓取网络,嵌入了提出的采样和学习方法,能够有效提升物体抓取检测性能。结果 在大型基准数据集GraspNet-1Billion上进行评估,本文方法取得对比方法中的最优性能,其中在小物体上的抓取指标平均提升了7%,大量的真实机器人实验也表明该方法具有抓取未知物体的良好泛化性能。结论 本文聚焦于小物体上的抓取,提出了一种掩码辅助采样方法嵌入到提出的端到端学习网络中,并引入了多尺度分组学习策略提高物体的局部几何表示,能够有效提升在小尺寸物体上的抓取质量,并在所有物体上的抓取评估结果都超过了对比方法。     

自然场景中的视觉显著对象检测

显著对象检测是视觉注意机制的一个重要应用基础研究,对于图像检索、场景分析、图像标注与对象识别都有着重要的研究意义。基于Tresiman特征整合理论和Koch计算框架,提出一种自然场景中视觉显著对象的检测方法。该方法首先建立适用于彩色自然场景的视觉显著度模型,计算多种不同特征的显著度,然后在融合不同特征的综合显著度图中提取显著对象。实验结果表明,与经典的Itti模型相比,这种方法不仅检测快速而且更准确地将视觉显著对象从背景中分离出来,更符合人眼的真实视觉注意过程。     

Independent motion detection in 3D scenes

This paper presents an algorithmic approach to the problem of detecting independently moving objects in 3D scenes that are viewed under camera motion. There are two fundamental constraints that can be exploited for the problem: 1) two/multiview camera motion constraint (for instance, the epipolar/trilinear constraint) and 2) shape constancy constraint. Previous approaches to the problem either use only partial constraints, or rely on dense correspondences or flow. We employ both the fundamental constraints in an algorithm that does not demand a priori availability of correspondences or flow. Our approach uses the plane-plus-parallax decomposition to enforce the two constraints. It is also demonstrated that for a class of scenes, called sparse 3D scenes in which genuine parallax and independent motions may be confounded, how the plane-plus-parallax decomposition allows progressive introduction, and verification of the fundamental constraints. Results of the algorithm on some difficult sparse 3D scenes are promising.