基于面向对象分类的芒果林遥感提取方法研究

发布时间：2018-07-02 15:04

本文选题：芒果林地 + 面向对象分类　；参考：《资源科学》2017年08期

【摘要】：中国芒果种植面积居世界第二,并有持续增长的趋势。国内外利用遥感手段提取农作物的相关研究较多,但有关芒果林遥感提取的研究仍较少。本研究基于2016年12月的高分辨率卫星SPOT-6数据,结合植被覆盖度(FVC)和坡度(SLOPE)因子,利用面向对象分类方法对芒果林地信息进行提取,结果表明,利用FVC和SLOPE参与分割的面向对象分类方法,提取芒果林地的生产者精度达92.81%,用户精度达97.19%。该方法相比于最大似然法分类以及FVC和SLOPE未参与分割的面向对象分类,提取芒果林地的生产者精度分别提高了19.28%和3.62%;用户精度分别提高了8.62%和3.86%。本研究可为果园用地的遥感识别与信息提取有效的方法借鉴。
[Abstract]:China has the second largest mango planting area in the world and has a growing trend. There are many researches on crop extraction by remote sensing at home and abroad, but there are few researches on remote sensing extraction of mango forest. Based on the high resolution satellite SPOT-6 data from December 2016 and combining the vegetation coverage (FVC) and slope (slope) factors, the information of mango forest land is extracted by using object-oriented classification method. The results show that, By using the object oriented classification method of FVC and slope, the precision of mango forest is 92.81 and 97.19.1 respectively. Compared with the maximum likelihood classification and the object oriented classification without FVC and slope, the accuracy of mango forest is improved by 19.28% and 3.62%, respectively, and the user accuracy is increased by 8.62% and 3.86% respectively. This study can be used for reference in remote sensing identification and information extraction of orchard land.
【作者单位】：中国科学院遥感与数字地球研究所数字地球重点实验室;中国科学院大学;海南省地球观测重点实验室;
【基金】：海南省自然科学基金项目(2016CXTD015) 海南省应用技术研发与示范推广专项(ZDXM2015102) 海南省重大科技计划项目(ZDKJ2016021-02)
【分类号】：S127;S667.7

【相似文献】