| 本文已被：浏览 1110次 下载 435次 |
码上扫一扫！ |
|
|
| 基于纹理特征与MODIS NDVI时间序列的耕地面积提取研究 |
|
于利峰1,2, 乌兰吐雅1,2, 乌云德吉1,2, 许洪滔1,2, 包珺玮1,2, 任婷婷1,2
|
|
1.内蒙古自治区农牧业科学院农牧业经济与信息研究所，呼和浩特010031;2.内蒙古自治区农业遥感工程技术研究中心，呼和浩特 010031
|
|
| 摘要: |
| ［目的］耕地是粮食安全的重要前提与保障，通过对MODIS时间序列数据的特征分析和提取，讨论了在大尺度条件下的耕地面积提取的可行方法，以期为当地合理利用耕地资源进行农业生产规划与布局提供参考。［方法］文章以呼伦贝尔市为研究区域，选择MOD13Q1为数据源，采用Savizky Golay 滤波方法对MODIS NDVI时间序列进行滤波处理，并分析各地物间的时间变化特征，最后结合纹理特征信息进行CART决策树分类的结果比较。［结果］（1）MODIS NDVI时间序列能较好的区分不同的土地利用类型； （2）Savizky Golay滤波降噪能够明显提高分类精度； （3）纹理特征的加入进一步反映地物间差异性； （4）该方法的总体分类精度为8372%，Kappa系数为0789，其中耕地的提取准确度为8633%。［结论］纹理特征使数据像元间的灰度相关性更加丰富，并通过提高像元之间的差异改善结果的精度。该文结果为进一步利用MODIS数据进行土地利用信息与农业资源调查提取提出了新思路。 |
| 关键词: 粮食安全MODIS数据纹理特征Savizky Golay滤波决策树 |
| DOI： |
| 分类号:S127 |
| 基金项目:内蒙古自然科学基金项目“基于高光谱的大兴安岭北麓主要农作物识别遥感机理研究”（2016MS（LH）0301）； 内蒙古农牧业科学院创新基金项目“基于北斗精准时空服务的大尺度农作物遥感识别方法研究”（2018CXJJN09）； 内蒙古财政厅项目“内蒙古主要农作物长势、估产遥感监测预报服务平台建设” |
|
| STUDY ON EXTRACTION OF ARABLE LAND AREA BASED ON TEXTURE FEATURES AND MODIS NDVI TIME SERIES |
|
Yu Lifeng1,2, Wulan Tuya1,2, Wuyun Deji1,2, Xu Hongtao1,2, Bao Junwei1,2, Ren Tingting1,2
|
|
1. Institute of Agricultural and Animal Husbandry Economy and Information, Inner Mongolia Academy of Agricultural and Animal Husbandry Sciences, Hohhot, 010031, China;2. Inner Mongolia Engineering and Technology Research Center for Agricultural Remote Sensing, Hohhot, 010031, China
|
| Abstract: |
| Arable land is an essential condition and guarantee of food security. Basing on analysis and extract from characteristics of MODIS time series data, the feasible method of extracting the arable area under large scale condition is discussed, so as to provide reference for agricultural planning in utilization of arable area resources. Took Hulun Buir city as the research area, selected MOD13Q1 as data source, the study adopted Savizky Golay filtering method to filter the MODIS NDVI time series and the time variation characteristics between different objects were analyzed. Lastly, the results of CART decision tree classification with texture feature information were compared. By comparison, the different classification results show that（1）The MODIS NDVI time series data can better distinguish different land use types;（2）The Savizky Golay filter can improve the accuracy of classification;（3）The addition of texture features further reflects the differences between the landmark;（4）The overall classification accuracy of this method is 83.72%,kappa coefficient is 0.789, in which the extraction accuracy of cultivated land is 86.33%. Base on the results, texture features enrich the gray correlation between the data cells and improve the accuracy of the results by improving the difference between the pixels. The study provide a new idea for the further use of MODIS data for land use information and agricultural resources investigation and extraction. |
| Key words: food security MODIS data texture features Savizky Golay filtering decision tree |
