遥感图像分类技术研究

Remotely Sensed Image Classification Technology Review

目的 探讨遥感图像分类技术 ,研究不同工作要求的分类方法 .方法 比较了目前流行的几种像分类算法 ,讨论了这些算法的特点 .结果 对实际遥感图像分析研究提供了工作方法 .结论 对快速遥感数据分析场合 ,基于一阶灰度统计的方法的结果是可以接受的 ;非实时和较高分类精度要求的场合 ,应考虑采用神经网络。

Aim\ Analyze framework of remotely sensed image classification technology to find the classification algorthms suitable for different requirement. Methods\ Compare several prevailing classification algorithms and describe their characteristic. Results\ Give the suitable classification algorithm for our application. Conclusion\ For the real time RS image applications, First order gray histogram based statistical methods can give quite good result, but for more accurate classification requirement, Neural network or Markov random field or other advanced methods should be considered.