Textural and spectral features as an aid to cloud classification

Abstract

The problem of classifying clouds seen on meteorological satellite images into different types is one which requires the use of textural as well as spectral information. Since multi-spectral features are of prime importance, textural features must be considered as augmenting, rather than replacing, spectral measures. Several textural features are studied to determine their discriminating power across a number of cloud classes including those which have previously been found difficult to separate. Although several features in the frequency domain are tested they are found to be less useful than those in the spatial domain with only one exception. The specific features recommended for use in classification depend on the type of classification to be undertaken. Specifically, different features should be used for a multi-dimensional feature space analysis than for a binary-tree rule-based classification.     

Temporal updating scheme for probabilistic neural network withapplication to satellite cloud classification

In cloud classification from satellite imagery, temporal change in the images is one of the main factors that causes degradation in the classifier performance. In this paper, a novel temporal updating approach is developed for probabilistic neural network (PNN) classifiers that can be used to track temporal changes in a sequence of images. This is done by utilizing the temporal contextual information and adjusting the PNN to adapt to such changes. Whenever a new set of images arrives, an initial classification is first performed using the PNN updated up to the last frame while at the same time, a prediction using Markov chain models is also made based on the classification results of the previous frame. The results of both the old PNN and the predictor are then compared. Depending on the outcome, either a supervised or an unsupervised updating scheme is used to update the PNN classifier. Maximum likelihood (ML) criterion is adopted in both the training and updating schemes. The proposed scheme is examined on both a simulated data set and the Geostationary Operational Environmental Satellite (GOES) 8 satellite cloud imagery data. These results indicate the improvements in the classification accuracy when the proposed scheme is used.     

Polarized visible light as an aid to vegetation classification

Radiation, when reflected from the surface of the earth, can be described in terms of both its radiance and its polarization and yet remote sensing has concerned itself with the measurement of radiance and has paid little attention to the measurement of polarization. However, the use of polarization measurements in remote sensing may increase as NASA have included polarizing filters on the satellite-borne Multispectral Resource Sampler (MRS), which may be launched in the mid-1980s. Photographic measurements of percent reflected visible light (percent RVL) and percent polarised visible light (percent PVL) were taken from a light aircraft on two summer days and two winter days. The study area was a heathland with seven land cover classes. In the summer, percent RVL, percent PVL, and percent RVL plus percent PVL could discriminate four land-cover classes. In the winter percent RVL plus percent PVL could discriminate five land-cover classes, percent PVL could discriminate four land-cover classes and percent RVL could discriminate only three land-cover classes. It was concluded that measurements of percent PVL when combined with measurements of percent RVL improved vegetation discrimination in winter months.     

The transportation algorithm as an aid to chromosome classification

An algorithm is presented which obtains a constrained maximum likelihood classification of homogeneously stained chromosomes. Significantly improved results over a both context-free and a plausible context-driven classification are obtained. Extension to banded chromosomes and abnormal cells are discussed.     

Temporal updating scheme for probabilistic neural network withapplication to satellite cloud classification - further results

A novel temporal updating approach for probabilistic neural network classifiers was developed by Tian et al. (2000) to account for temporal changes of spectral and temperature features of clouds in the visible and infrared GOES 8 (Geostationary Operational Environmental Satellite) imagery data. In this paper, a new method referred to as moving singular value decomposition (MSVD) is introduced to improve the classification rate of the boundary blocks or blocks containing cloud types with non-uniform texture. The MSVD method is then incorporated into the temporal updating scheme and its effectiveness is demonstrated on several sequences of GOES 8 cloud imagery data. These results indicate that the incorporation of the new MSVD improves the overall performance of the temporal updating process by almost 10%     

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.     

高分辨卫星图像卷积神经网络分类模型

目的 卫星图像往往目标、背景复杂而且带有噪声,因此使用人工选取的特征进行卫星图像的分类就变得十分困难。提出一种新的使用卷积神经网络进行卫星图像分类的方案。使用卷积神经网络可以提取卫星图像的高层特征,进而提高卫星图像分类的识别率。方法 首先,提出一个包含六类图像的新的卫星图像数据集来解决卷积神经网络的有标签训练样本不足的问题。其次,使用了一种直接训练卷积神经网络模型和3种预训练卷积神经网络模型来进行卫星图像分类。直接训练模型直接在文章提出的数据集上进行训练,预训练模型先在ILSVRC（the ImageNet large scale visual recognition challenge）-2012数据集上进行预训练,然后在提出的卫星图像数据集上进行微调训练。完成微调的模型用于卫星图像分类。结果 提出的微调预训练卷积神经网络深层模型具有最高的分类正确率。在提出的数据集上,深层卷积神经网络模型达到了99.50%的识别率。在数据集UC Merced Land Use上,深层卷积神经网络模型达到了96.44%的识别率。结论 本文提出的数据集具有一般性和代表性,使用的深层卷积神经网络模型具有很强的特征提取能力和分类能力,且是一种端到端的分类模型,不需要堆叠其他模型或分类器。在高分辨卫星图像的分类上,本文模型和对比模型相比取得了更有说服力的结果。     

基于时空优化深度神经网络的AQI等级预测

针对现有空气质量预测方法精度偏低、对噪声敏感等问题,提出一种基于堆栈降噪自编码（Stacked Denoising Auto-Encoders,SDAE）模型的空气质量等级预测方法。首先以武汉市历史空气质量和气象监测数据为研究对象,建立SDAE模型逐层学习原始数据的特征表达,并将最后一层特征与分类器连接完成预测模型的调优。同时改进多参数网格搜索法,选取了最优的超参数组合。然后在测试集上进行预测,并用预测值与实际值之间的平均绝对误差和均方误差等指标作为预测性能评价标准。通过与其他网络模型的实验对比,证明了SDAE模型对于空气质量等级具有较优的预测性能。最后从时间、空间、时空三个角度对该模型输入进行优化,实验结果表明基于空间优化的SDAE模型预测性能提升最为明显,能够得到比传统方法更加精确的预测结果。     

A neural network approach to the classification problem

The task of classifying observations into known groups is a common problem in decision making. A wealth of statistical approaches, commencing with Fisher's linear discriminant function, and including variations to accomodate a variety of modeling assumptions, have been proposed. In addition, nonparametric approaches based on various mathematical programming models have also been proposed as solutions. All of these proposed aolutions have performed well when conditions favorable to the specific model are present. The modeler, therefore, can usually be assured of a good solution to his problem of he chooses a model which fits his situation. In this paper, the performance of a neural network as a classifier is evaluated. It is found that the performance of the neural network is comparable to the best of otheother methods under a wide variety of modeling assumptions. The use of neural networks as classifiers thus relieves the modeler of testing assumptions which would otherwise be critical to the performance of the usual classification techniques.     

Respiratory sound classification by using an incremental supervised neural network

Determination of lung condition by auscultation is a difficult task and requires special training of medical staff. It is, however, a difficult skill to acquire. In decision making, it is significant to analyze respiratory sounds by an algorithm to give support to medical doctors. In this study, first, a rectangular window is formed so that one cycle of respiratory sound (RS) is contained in this window. Then, the windowed time samples are normalized. In order to extract the features, the normalized RS signal is partitioned into 64 samples of long segments. The power spectrum of each segment is computed, and synchronized summation of power spectra components is performed. Feature vectors are formed by the averaged power spectrum components, yielding 32-dimensional vectors. In the study, classification performances of multi-layer perceptron (MLP), grow and learn (GAL) network and a novel incremental supervised neural network (ISNN) are comparatively examined for the classification of nine different RS classes: Bronchial sounds, bronchovesicular sounds, vesicular sounds, crackles sounds, wheezes sounds, stridor sounds, grunting sounds, squawk sounds, and sounds of friction rub.     

基于改进卷积神经网络的多标记分类算法

良好的特征表达是提高模型性能的关键,然而当前在多标记学习领域,特征表达依然采用人工设计的方式,所提取的特征抽象程度不高,包含的可区分性信息不足。针对此问题,提出了基于卷积神经网络的多标记分类模型ML_DCCNN,该模型利用卷积神经网络强大的特征提取能力,自动学习能刻画数据本质的特征。为了解决深度卷积神经网络预测精度高,但训练时间复杂度不低的问题,ML_DCCNN利用迁移学习方法缩减模型的训练时间,同时改进卷积神经网络的全连接层,提出双通道神经元,减少全连接层的参数量。实验表明,与传统的多标记分类算法以及已有的基于深度学习的多标记分类模型相比,ML_DCCNN保持了较高的分类精度并有效地提高了分类效率,具有一定的理论与实际价值。     

Wetland classification using optical and radar data and neural network classification

A study was conducted to investigate the ability of a neural network based classification technique to delineate upland and forested wetland areas and to distinguish between different levels of wetness in a forested wetland. NASA's Airborne Terrestrial Applications Sensor (ATLAS) multi-spectral data and Airborne Imaging Radar Synthetic Aperture Radar (AIRSAR) data were used in this study. A National Wetland Inventory (NWI) map served as a reference. Cascade-correlation, a feed-forward neural network architecture, was employed as the classifier. The neural network technique separated upland from wetland spectral signatures and discriminated two out of four different water regimes identified by the NWI within the wetland area. The relative usefulness of ATLAS and AIRSAR data for wetness classification was also investigated. It was found that both data sources, when used in isolation, could separate wetland from upland about equally well, but better performance was observed when these data sources were combined.     

File-aware P2P traffic classification: An aid to network management

As P2P dominates Internet traffic in recent years, ISPs are striving to balance between providing the basic networking services for P2P users and properly managing network bandwidth usage. That is, ISPs are required to provide proper bandwidth for each P2P user to get every file to fulfill their provision for communications, while they have to control bandwidth consumption for efficient usage. However, current P2P traffic management strategies are unable to satisfy both requirements. In this paper, our goal is to design a simple and effective scheme for ISPs to moderate the tradeoff. It is achieved by proposing a file-aware P2P traffic classification method that can identify files and the associated flows. The file-level information can lead to more efficient and flexible management strategies on a per-file basis. We offer two alternatives: constraining the per-file bandwidth consumption and the number of per-file concurrent flows. Finally, a real-life trace is measured using our file-aware method from the perspectives of peers and files. The results indicate that ISPs can gain enough opportunities to flexibly choose proper traffic manage parameters according to actual demands.     

SDANet: spatial deep attention-based for point cloud classification and segmentation

Machine Learning - Using deep learning to learn point cloud features directly have become one of the research hotspots in the field of 3D point cloud processing. The existing methods usually...     

基于神经网络的电子邮件分类与过滤

现在电子邮件的应用非常广泛,已经成为人们生活中一种重要的通讯手段,但各种各样的垃圾邮件也是令我们十分困扰的问题,给出了一种电子邮件的分类过滤方法。电子邮件作为一种半结构化的文档,电子邮件信息包含了固定的语法部分和一定长度的可变文本部分,同时处理这两部分以得到更高的准确度。首先对邮件进行文本处理,得到特征向量;然后使用基于神经网络的方法对邮件进行分类过滤得到邮件分类器;最后通过实验验证分类器的有效性。     

一种概率过程神经元网络模型及分类算法

针对动态信号分类及与先验类别知识融合问题,提出了一种概率过程神经元网络模型．模型将贝叶斯概率分类机制与过程神经元网络动态信号处理方法相结合,通过在前馈过程神经元网络中增加一个模式单元层,以及采用归一化指数类型激励函数,实现基于贝叶斯规则的动态信号分类．分析了概率过程神经元网络分类机制与贝叶斯分类规则的等价性,给出了具体的学习算法,实验结果验证了模型和算法的有效性．     

Development of an operational cloud classification model

Abstract

Within the PROgramme for a Meteorological Information System (PROMIS) project in Sweden a method of multi-spectral analysis and classification of Advanced Very High Resolution Radiometer (AVHRR) data is developed. The model will be based on a statistical database of object classes and ruled by information about prevailing Sun elevations and air mass temperatures. A new system provides the image processing and data handling capacity necessary for an operational classification of AVHRR data. Operational tests will start during the beginning of 1988.     

Enhanced brain tumor classification using an optimized multi-layered convolutional neural network architecture

Multimedia Tools and Applications - Detecting and classifying a brain tumor is a challenge that consumes a radiologist’s time and effort while requiring professional expertise. To resolve...     

基于时空循环神经网络的下一个兴趣点推荐方法

下一个兴趣点推荐已经成为基于位置的社交网络(location-based social networks,LBSNs)中一个重要任务.现有的模型没有深入考虑相邻签到兴趣点之间的转移时空信息,无法对用户访问下一个兴趣点的长短时间偏好和远近距离偏好进行有效建模.本文通过对循环神经网络(recurrent neural ne...