基于SPOT-5卫星影像的水稻信息提取方法研究——以德阳市旌阳区为例
发布时间：2018-10-08 07:41
【摘要】：【目的】水稻遥感信息提取是遥感技术在农业领域应用方面的重要内容,也是快速、准确满足水稻种植遥感监测的需要。【方法】本研究以四川省德阳市旌阳区为研究区,利用SPOT-5卫星影像,对研究区的影像进行监督、面向对象以及决策树等多种方法分类,对分类结果进行对比,研究最适合提取水稻信息的方法。【结果】结果表明:(1)监督分类(6种分类器)人为控制训练区提高精度的同时也加大了人为误差;面向对象分类提高了效率,易出现分类混淆;决策树分类法直观、效率高,但在本研究区中,由于耕林混交的面积较大,水体和居民地亮度值接近,造成分类误差加大。(2)神经网络和支持向量机的分类精度最高,分类效果清晰,说明在实际水稻信息提取中以监督分类为最佳。【结论】基于遥感技术和高分辨率数据提取水稻信息、实现水稻监测是可行的。
[Abstract]:[objective] Rice remote sensing information extraction is an important content in the application of remote sensing technology in the agricultural field, and it is also rapid and accurate to meet the needs of remote sensing monitoring of rice cultivation. [methods] in this study, Jingyang District, Deyang City, Sichuan Province, was taken as the research area. The SPOT-5 satellite image is used to monitor the image of the research area, object oriented and decision tree are used to classify the image, and the classification results are compared. The results show that: (1) the artificial control training area of supervised classification (6 classifiers) improves the accuracy and increases the artificial error, and the object-oriented classification improves the efficiency and is liable to be confused; Decision tree classification method is intuitionistic and efficient, but in this study area, the brightness value of water body and residents is close, which results in the higher classification error. (2) the classification accuracy of neural network and support vector machine is the highest. The classification effect is clear, which shows that the best way to extract rice information is supervised classification. [conclusion] it is feasible to extract rice information based on remote sensing technology and high-resolution data.
【作者单位】： 四川师范大学地理与资源科学学院;四川师范大学西南土地资源评价与监测教育部重点实验室;四川省农业科学院遥感应用研究所;
【基金】：国家自然科学基金项目“基于LUCC扰动影响的成都平原土地生态安全维持机理”(41371125)
【分类号】：S127;S511
本文编号：2255967
[Abstract]:[objective] Rice remote sensing information extraction is an important content in the application of remote sensing technology in the agricultural field, and it is also rapid and accurate to meet the needs of remote sensing monitoring of rice cultivation. [methods] in this study, Jingyang District, Deyang City, Sichuan Province, was taken as the research area. The SPOT-5 satellite image is used to monitor the image of the research area, object oriented and decision tree are used to classify the image, and the classification results are compared. The results show that: (1) the artificial control training area of supervised classification (6 classifiers) improves the accuracy and increases the artificial error, and the object-oriented classification improves the efficiency and is liable to be confused; Decision tree classification method is intuitionistic and efficient, but in this study area, the brightness value of water body and residents is close, which results in the higher classification error. (2) the classification accuracy of neural network and support vector machine is the highest. The classification effect is clear, which shows that the best way to extract rice information is supervised classification. [conclusion] it is feasible to extract rice information based on remote sensing technology and high-resolution data.
【作者单位】： 四川师范大学地理与资源科学学院;四川师范大学西南土地资源评价与监测教育部重点实验室;四川省农业科学院遥感应用研究所;
【基金】：国家自然科学基金项目“基于LUCC扰动影响的成都平原土地生态安全维持机理”(41371125)
【分类号】：S127;S511
【相似文献】
相关期刊论文 前2条
1 钱永兰,杨邦杰,雷廷武;用基于IHS变换的SPOT-5遥感图像融合进行作物识别[J];农业工程学报;2005年01期
2 ;[J];;年期
，本文编号：2255967
本文链接：https://www.wllwen.com/kejilunwen/nykj/2255967.html
