论文部分内容阅读
仅依靠光谱信息无法满足高分辨率遥感分类的应用需求,辅之以纹理特征信息进行分类,可提高影像分类精度。利用KZ-1卫星影像和Landsat-8卫星影像数据,基于面向对象的影像分割法和灰度共生矩阵纹理分析法对新疆石河子市局部城区进行了地表覆盖分类实验,将不同空间分辨率的全色影像纹理信息、光谱信息构成多种影像特征组合进行分类比较研究,以选择最佳的分类特征集。结果表明:KZ-1影像能为城市区域的土地覆盖分类提供丰富的纹理信息,面向对象的影像分割可较好地利用高分辨率数据的几何结构信息实现优化的影像分割,从而提高多光谱影像的分类精度,总体分类精度为90.06%,Kappa系数为87.93%,比单纯利用光谱信息分类的总体精度提高了8.02%,Kappa系数提高了9.65%,表明KZ-1数据可为光谱分类提供丰富的纹理信息,从而提高城市区域的土地覆盖分类精度。
Only rely on the spectral information can not meet the application requirements of high resolution remote sensing classification, combined with the texture feature information classification, can improve the image classification accuracy. Using KZ-1 satellite imagery and Landsat-8 satellite imagery, based on the object-oriented image segmentation method and the gray level co-occurrence matrix texture analysis method, the surface coverage classification experiment was carried out in the local urban areas of Shihezi, Xinjiang. The total spatial resolution Image texture information and spectral information constitute a variety of image feature combinations for classification and comparative study to select the best classification feature set. The results show that KZ-1 image can provide abundant texture information for land cover classification in urban areas. Object-oriented image segmentation can make use of the geometric structure information of high-resolution data to achieve optimal image segmentation and improve the multi-spectral image , The overall classification accuracy was 90.06% and the Kappa coefficient was 87.93%, which increased by 8.02% and Kappa coefficient increased by 9.65% compared with that of spectral information alone, which indicated that KZ-1 data could provide abundant Texture information, thereby improving the accuracy of land cover classification in urban areas.