华中科技大学学报（自然科学版）2011，Vol.39Issue(z2)：113-115,3.
面向对象的遥感图像水陆场景快速分类
Water-land scene classification of remote sensing image using object-oriented method
摘要
Abstract
An efficient method for water-land scene classification was proposed, which can be divided into three steps. In the first step, extract the texture features of the image for clustering and a coarse segmentation result is obtained. Secondly, merge the regions of the former result. In the final step, a measurement called edge density is defined as a feature for regions. By statistically analyzing plenty of samples, we determine an empirical value as the classification threshold. Through the comparison with traditional method, the experiment demonstrated that the proposed approach can improve the efficiency of the algorithm highly under the precondition of semantic accuracy. As a result, the fast scene classification was realized.关键词
水陆场景/分类/面向对象/纹理特征/区域合并Key words
water-land scene/ classification/ object-oriented/ texture feature/ region merging分类
信息技术与安全科学引用本文复制引用
王岳环,唐为林,王军..面向对象的遥感图像水陆场景快速分类[J].华中科技大学学报（自然科学版）,2011,39(z2):113-115,3.基金项目
国家自然科学基金资助项目(60736010) （60736010）
航天科技创新基金资助项目(CASC200902) （CASC200902）
国防预研基金资助项目(9140A010609JW0514). （9140A010609JW0514）
