A novel change detection approach based on visual saliency and random forest from multi-temporal high-resolution remote-sensing images

ABSTRACT

This article presents a novel change detection (CD) approach for high-resolution remote-sensing images, which incorporates visual saliency and random forest (RF). First, highly homogeneous and compact image super-pixels are generated using super-pixel segmentation, and the optimal segmentation result is obtained through image superimposition and principal component analysis. Second, saliency detection is used to guide the search of interest regions in the initial difference image obtained via the improved robust change vector analysis algorithm. The salient regions within the difference image that correspond to the binarized saliency map are extracted, and the regions are subject to the fuzzy c-means (FCM) clustering to obtain the pixel-level pre-classification result, which can be used as a prerequisite for super-pixel-based analysis. Third, on the basis of the optimal segmentation and pixel-level pre-classification results, different super-pixel change possibilities are calculated. Furthermore, the changed and unchanged super-pixels that serve as the training samples are automatically selected. The spectral features and Gabor features of each super-pixel are extracted. Finally, super-pixel-based CD is implemented by applying RF based on these samples. Experimental results on Quickbird, Ziyuan 3 (ZY3), and Gaofen 2 (GF2) multi-spectral images show that the proposed method outperforms the compared methods in the accuracy of CD, and also confirm the feasibility and effectiveness of the proposed approach.     

Object-oriented change detection based on the Kolmogorov–Smirnov test using high-resolution multispectral imagery

In this article, we propose a novel method of object-oriented change detection for high-resolution remote-sensing imagery. The method consists of three main parts: image segmentation, object adjusting and change detection. We use the Fractal Net Evolution Approach to segment the multi-temporal images. Then we adjust the object maps. By merging the objects in relatively large areas, the object -adjusting algorithm aims to obtain a set of objects with different sizes, which coincide better with the real ground objects than the single-scale results. In the third part, the Kolmogorov–Smirnov two-sample test detects each pair of objects in the multi-temporal object maps with multi-scale. The calculated value of the D-statistic is compared to the threshold of a user-defined significance level. Through these three processes, we can make full use of the spatial and spectral features in high- resolution images to detect changes. According to our experiments in two study areas employing QuickBird imagery, the overall errors of our method decreased by more than 1000 pixels compared with the conventional object-oriented change vector analysis. The proposed method can also avoid the errors resulting from classification in the method of post-classification comparison.     

Using real-time SAR simulation to assist pattern recognition applications in urban areas

In this paper, the impact of current developments in urban modeling and graphics programming to the potential of SAR simulation and interpretation in urban areas is presented. As it will be demonstrated, SAR simulation is feasible in real-time even for complex urban environments by applying modern graphics cards. For this purpose, the SAR geometry is implemented using programmable graphics processing units (GPUs), which are available as standard graphics hardware for 3D acceleration. Originally, the geometric models of urban areas, which provide the database for SAR simulation, were captured to generate realistic visualizations of virtual scenes. As will be demonstrated, such area covering databases can be modified to meet specific requirements of SAR simulations at different scales. Finally, the applicability of our SAR simulation as made available from off-the-shelf hardware components for the analysis and interpretation of SAR scenes by pattern recognition will be demonstrated.     

电力物联网的风险分析及安全措施研究

电力物联网覆盖了智能电网的状态感知、信息传输及智能处理各个环节,全面提高了电网的信息感知深度、广度以及密度。但是,由于网络应用模式的演变、应用环境的复杂化以及新的网络技术的广泛应用,给电力物联网带来新的安全问题。文章深入分析和研究电力物联网特征及各个层次面临的安全威胁和攻击模式,提出了相应的安全防护措施,给出了电力物联网安全防护架构及其体系,为电力物联网安全提供有力支撑。     

A multi-agent model system for land-use change simulation

This paper presents a multi-agent model system to characterize land-use change dynamics. The replicable parameterization process should be useful to the development of simulation frameworks, important to environmental policy makers to analyze different scenarios during decision making process. The methodological two-fold approach intends to form a solid backbone based on: (i) the systematic and structured empirical characterization of the model; and (ii) the conceptual structure definition according to the agent-based model documentation protocol – Overview, Design concepts and Details. A multi-agent system for land-use change simulation was developed to validate the model, which is illustrated with a case study of the Brazilian Cerrado using LANDSAT ETM images. The simulation results prove the model importance with a figure of merit greater than 50%, what means the amount of correctly predicted change is larger than the sum of any type of error. The results are very good compared with nine popular peer-reviewed land change models.     

A heterogeneous multi-criteria multi-expert decision-support system for scoring combinations of flood mitigation and recovery options

In this study, we developed an innovative operational decision-support system (DSS) based on flood data and mitigation or recovery options, that can be used by both naïve and expert users to score portfolios of flood mitigation or recovery measures. The DSS combines exposure (i.e., economic, social, or environmental values at risk) and resilience (i.e., protection of the main equilibrium functions of human and physical systems). Experts from different fields define indices and functions, stakeholders express their attitudes towards risk, relative weights, and risk perceptions, and both groups use a shared learning process for risk assessment. The DSS algorithms include the “technique for order performance by similarity to ideal solution” (TOPSIS) and the “basic linguistic term set” (BLTS) methods for heterogeneous multi-criteria multi-expert decision-making. Decisions are illustrated using fixed or bounded values of flood depth, duration, and frequency, with plausible parameter values, for a case study of Cesenatico. The best mitigation option was construction of sand dunes and development of evacuation plans, which achieved 32% of the potential net benefit. The best recovery option was construction of sand dunes and development of evacuation plans and insurance schemes, which achieved 42% of the potential net benefit. Mitigation options outperformed recovery options whenever the relative importance of exposure with respect to resilience was greater than 95%. Sensitivity analysis revealed that the best mitigation option was most robust with respect to flood duration and depth; the best recovery option was most robust with respect to the relative weights attached to economic, social, and environmental factors. Both options were similarly robust with respect to interdependencies between the options.     

基于小波域TS-MRF模型的监督图像分割方法

定义在单一空间分辨率上的树结构马尔可夫场(Tree-Structured Markov Random Field,TS-MRF)模型能够表达图像的分层结构信息,但难以描述图像的非平稳性,针对该问题,提出小波域的TS-MRF图像建模方法-WTS-MRF模型,按照图像分类层次树的结构形式,该模型将一系列的MRF嵌套定义在多...     

Simple multiscale algorithm for layer detection with lidar

Lidar is a powerful active remote sensing device used in the detection of the optical properties of aerosols and clouds. However, there are difficulties in layer detection and classification. Many previous methods are too complex for large dataset analysis or limited to data with too high a signal-to-noise ratio (SNR). In this study, a mechanism of multiscale detection and overdetection rejection is proposed based on a trend index function that we define. Finally, we classify layers based on connected layers employing a quantity known as the threshold of the peak-to-base ratio. We find good consistency between retrieved results employing our method and visual analysis. The testing of synthetic signals shows that our algorithm performs well with SNRs higher than 4. The results demonstrate that our algorithm is simple, practical, and suited to large dataset applications.     

Feature Selection with Data Field

A new feature selection method is proposed for high-dimensional data clustering on the basis of data field. With the potential entropy to evaluate the importance of feature subsets, features are filtered by removing unimportant features or noises from the original datasets. Experiments show that the proposed method can sharply reduce the number of dimensions and effectively improve the clustering performance on WDBC dataset.