论文部分内容阅读
在遥感中,所谓的变化即地表组分随时间的推移而发生的改变。土地覆盖和土地利用的变化信息是极其重要的,它们被广泛地应用到森林开采、灾害监测、灾害评估、城市扩张和土地管理等行业中。变化检测的工作流程即使用多时相数据集来定性地分析这些现象和对象的时序影响并量化其变化。遥感数据因具有较高的时间频率、便于计算的数据格式、天空视角、广阔的空间和光谱分辨率范围,已成为一个主要的变化检测数据源。在遥感中变化检测的主要目标有定义图像的几何位置和变化的类型,定量描述变化的大小以及评价变化检测结果的精度。传统的变化检测方法为基于像素的方法,而随着高空间分辨率影像的出现,面向对象的方法和数据挖掘技术被越来越广泛地应用到遥感变化检测当中。本文介绍了这三种方法并比较了它们的优缺点。
In remote sensing, the so-called change is the change of surface components over time. Information on land cover and land-use change is of paramount importance and is widely used in industries such as forest exploitation, disaster monitoring, disaster assessment, urban expansion and land management. The workflow of change detection uses multi-temporal datasets to qualitatively analyze the temporal effects of these phenomena and objects and quantify their changes. Remote sensing data has become a major source of change detection due to its high temporal frequency, easy-to-compute data format, sky view, wide spatial and spectral resolution range. The main targets of change detection in remote sensing are to define the geometric position and type of change of the image, to quantitatively describe the size of the change and to evaluate the accuracy of the change detection result. Traditional change detection methods are pixel-based methods. With the advent of high spatial resolution images, object-oriented methods and data mining techniques are increasingly used in remote sensing change detection. This article describes these three methods and compares their strengths and weaknesses.