基于SVM的环境减灾卫星HJ-1B影像作物分类识别研究

环境减灾卫星作为我国自主研制发射的环境与灾害监测预报卫星，要发挥它的作用就是要更好的使用其数据源。支持向量机（SupportVectorMachine，SVM）是一种卓越的分类方法，本文通过SVM方法对环境减灾卫星14J-1B星CCD影像数据进行作物分类识别实验并将结果与最大似然法分类结果进行比较。结果表明：利用SVM方法进行遥感图像分类，精度优于传统的最大似然法分类精度；HJ—1A／1B星CCD数据对于农作物具有较好的指示效果，可应用于作物识别等农业领域。

Study of Crop Classification by Support Vector Machine on HJ-1B Image

The Environment and Disaster Monitoring Microsatellite Constellation was put forward by China to achieve the large-area, All-weather, All-time dynamic monitoring of environment and disasters,and then to ensure the continuous stable development of national economy and society. Support Vector Machine (SVM) has excellent performance in classification. Study of crop classification by SVM on HJ-1B image of Envinmment and Disasters Monitoring Microsatellite show that:the accuracy obtained from SVM method is better than maximum likeli-hood;HJ-1A/1B Image of Environment and Disasters Monitoring Microsatellite cannot only be used for crop classification, but also achieve the best Classification accuracy.