基于多源高分卫星影像的果棉套种信息提取

发布时间：2018-07-05 05:55

本文选题：面向对象 + 多尺度分割　；参考：《国土资源遥感》2017年02期

【摘要】：棉花与果树间作在新疆多地区普遍存在,了解套种情况有利于查明果棉产量以及与常规棉田产量结构差异。为此,提出了一种综合使用多源高分遥感数据的果棉间作信息提取方法。首先,在优化分割尺度基础上分析Quick Bird卫星数据的光谱、形状和纹理特征并建立规则集;其次,使用面向对象的分类方法逐步剔除非农田信息形成地块专题图,基于专题图选择最佳纹理特征提取果树分布并以地块为单位统计套种比例;最后,依据棉花物候特征对高分一号数据多时相分类得到棉花种植信息,结合套种比例结果,统计果棉套种面积及程度。精度检验结果表明:该文提出的方法与传统抽样调查法相比能够为大量地块信息的采集节省人工成本和时间,果棉信息提取精度为89.16%,可以在统计调查工作中用于新疆果棉套种的自动化提取。
[Abstract]:The intercropping of cotton and fruit trees is common in many areas of Xinjiang. It is helpful to find out the yield of fruit and cotton and the difference of yield structure between cotton field and conventional cotton field. In this paper, a method for extracting intercropping information of cotton and fruit is proposed, which synthetically uses multi-source and high-score remote sensing data. Firstly, the spectral, shape and texture features of Quick Bird satellite data are analyzed on the basis of optimizing segmentation scale, and the rule set is established. Selecting the best texture feature based on thematic map to extract fruit tree distribution and statistical interplanting proportion based on plot; finally, according to cotton phenological characteristics to the high-score No. 1 data multi-temporal classification of cotton planting information, combined with the results of interplanting proportion, The area and degree of interplanting of fruit and cotton were counted. The accuracy test results show that the method proposed in this paper can save labor cost and time compared with the traditional sampling survey method. The precision of fruit cotton information extraction is 89.16, which can be used in the automatic extraction of interplanting in Xinjiang.
【作者单位】：中国地质大学(北京)土地科学技术学院;中国科学院遥感与数字地球研究所遥感科学国家重点实验室;
【基金】：国家自然科学基金“基于高分辨率逐日模拟遥感数据的农作物物候参数精确提取研究”(编号:41371358) 国家高技术研究发展计划(“863”计划)“先进环境监测技术设备——星-机-地生态环境质量遥感监测系统集成与示范”(编号:2014AA06A511)共同资助
【分类号】：S562;S66;TP751

【相似文献】