基于Web服务的高分辨率遥感影像在线分类技术研究

发布时间：2018-03-04 21:08

本文选题：在线分类　切入点：Web　出处：《中国矿业大学》2017年硕士论文　论文类型：学位论文

【摘要】：高分辨率遥感卫星的出现,使得遥感影像的应用越来越广泛,在农业、林业、牧业、国土、水利、环境、测绘、交通、灾害、气象、城乡建设等行业领域都发挥了重要作用。传统的遥感影像处理分析是通过专业人员在单机上实现的,随着网络技术的高速发展,如何以Web的方式向用户提供遥感数据的社会化服务成为一个研究的热点问题。本文就高分辨率遥感影像在线分类技术框架以及信息分析结果可视化实现展开研究,通过整合遥感数据处理算法及处理工具资源进行系统集成,并利用Web Service技术提供标准服务,提出了基于Web服务的影像在线分类技术的解决方案。论文具体的研究工作主要有:(1)详细地介绍了遥感影像在线服务的技术基础以及服务开发模式,利用ENVI Services Engine将自定义的专业遥感影像计算分析算法模型发布成标准REST服务供客户端调用,设计并实现了高分辨率遥感影像在线分类系统,提供给用户一种在线的、实时的、方便的专业化服务资源。(2)针对遥感数据在线服务出现的问题,提出了一种针对大数据量矢量在线快速加载的解决方案。结合Geo JSON技术实现了矢量轻量级存储及传输,通过网格化矢量显示区域,实现了矢量数据的按需显示,大大减少了矢量数据加载及渲染的时间,有效地解决了大数据量矢量绘制可能导致的浏览器崩溃的问题,并利用异步加载策略来增强用户的体验。(3)提出了一种基于多分类器混联模型的遥感影像分类算法。根据样本在整体最优分类器的表现情况调整类别最优分类器的级联顺序,完成自适应级联模型的构建;对于级联模型拒识的样本,利用改进的贝叶斯平均融合规则进行二次决策,输出最终分类结果。实验证明,本文方法能够有效地识别高分辨率遥感影像地物类别。
[Abstract]:The appearance of high-resolution remote sensing satellite makes the application of remote sensing image more and more extensive, in agriculture, forestry, animal husbandry, land, water conservancy, environment, mapping, transportation, disaster, meteorology, Traditional remote sensing image processing and analysis are realized on a single computer by professionals. With the rapid development of network technology, the traditional remote sensing image processing and analysis has played an important role in the field of urban and rural construction. How to provide users with social services of remote sensing data by Web has become a hot issue. This paper studies the technical framework of online classification of high-resolution remote sensing images and the realization of visualization of information analysis results. Through the integration of remote sensing data processing algorithms and processing tools resources, and the use of Web Service technology to provide standard services, This paper puts forward the solution of image online classification technology based on Web service. The main research work in this paper is: 1) the technical foundation and service development mode of remote sensing image online service are introduced in detail. Using ENVI Services Engine, the professional remote sensing image calculation and analysis algorithm model is published as a standard REST service for client to call. A high resolution remote sensing image online classification system is designed and implemented, which provides users with an online and real-time remote sensing image classification system. Aiming at the problems of remote sensing data online service, this paper puts forward a solution for fast loading of large data vector online, and combines Geo JSON technology to realize vector lightweight storage and transmission. Through gridding vector display area, vector data display on demand is realized, the time of vector data loading and rendering is greatly reduced, and the problem of browser crash caused by vector rendering with large amount of data is effectively solved. An algorithm of remote sensing image classification based on multi-classifier hybrid model is proposed by using asynchronous loading strategy to enhance the user's experience. The cascade order of class optimal classifier is adjusted according to the performance of the sample in the overall optimal classifier. The adaptive cascade model is constructed, and the improved Bayesian average fusion rule is used to make the quadratic decision for the rejected samples of the cascade model, and the final classification results are outputted. The method in this paper can effectively identify the ground objects in high resolution remote sensing images.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP751;TP393.09

【参考文献】