• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>American Society for Photogrammetry and Remote Sensing Annual Conference >IMPROVING CLASSIFICATION ACCURACY OF SPECTRALLY SIMILAR URBAN CLASSES BY USING OBJECT-ORIENTED CLASSIFICATION TECHNIQUES: A CASE STUDY OF NEW YORK CITY
【24h】

IMPROVING CLASSIFICATION ACCURACY OF SPECTRALLY SIMILAR URBAN CLASSES BY USING OBJECT-ORIENTED CLASSIFICATION TECHNIQUES: A CASE STUDY OF NEW YORK CITY

机译：采用面向对象的分类技术提高光谱相似的城市课程的分类准确性：纽约市的案例研究

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Paper describes a methodology to improve classification of urban features by using object oriented classification techniques. Mapping urban features from satellite data is challenging due to several reasons. Urban objects are spectrally similar and they have different shapes, sizes, patterns, all of which contribute to their low accuracy. For example, features such as roofs, roads, and other spectrally similar objects like open space (concrete) appear spectrally similar leading to their low accuracy in classification. Very high resolution satellite data (Ikonos) was classified using both supervised and object oriented classification techniques. A combination of spectral, spatial attributes and membership functions were employed for mapping urban features. Accuracy assessment was carried out using ground truth data acquired from field surveys and from other reliable secondary data sources. Whilst the per-pixel supervised classification approach produced reasonable overall accuracy, specific classes such as white roof and vegetation registered low user's accuracy (79.82% and 70.07%) respectively. These classes were mapped using an object oriented classification approach. Spectral thresholds and membership functions were employed after the image was segmented. Results show the potential benefits of using object oriented classification techniques for mapping urban features using VHR satellite data. The methodology and results have useful implications for improving urban mapping accuracy particularly for those data that have high spatial resolution but low spectral resolution. GIS layers may be extracted after image analysis that can assist in the production of high quality digital maps for planning, emergency applications, risk assessment, integration with census data, analysis and modeling.
机译：纸张介绍通过使用面向对象的分类技术来改善城市特征分类的方法。由于几个原因，绘制来自卫星数据的城市特征是挑战。城市对象是光谱相似的，它们具有不同的形状，尺寸，图案，所有这些都有助于它们的低精度。例如，屋顶，道路和其他频道类似物体等特征，如开放空间（混凝土）看起来具有光谱相似，导致它们在分类中的低精度。使用监督和面向对象的分类技术分类非常高的分辨率卫星数据（IKONOS）。频谱，空间属性和隶属函数的组合用于映射城市特征。使用从现场调查和其他可靠的二级数据源获取的地面真理数据进行准确性评估。虽然每个像素的监督分类方法产生合理的整体准确性，但白色屋顶和植被等特定类别分别登记了低用户的准确性（79.82％和70.07％）。使用面向对象的分类方法映射这些类。在分段图像后采用光谱阈值和隶属函数。结果表明使用VHR卫星数据使用面向对象的分类技术来映射城市特征的潜在好处。方法论和结果对于提高城市映射精度具有有用的影响，特别是对于具有高空间分辨率但低频分辨率的数据。可以在图像分析之后提取GIS层，可以帮助生产高质量的数字地图进行规划，应急应用，风险评估，与人口普查数据集成，分析和建模。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号