基于A-Clenet5的荒漠化草原草种识别与分类

草原退化已经成为人类面临的主要生态问题,其标 志之一是草原植被群落结构的改变,而草原草种的高光谱识别与分类是利用遥感进行大面积 高精度草原退化监测与治理的基础与前提。由于天然草原草种分布的随机性和遥感图像云系 情况的复杂性,草种识别精度低的问题未能得到满意的解决。本研究基于深度学习首次提出 A-Clenet5法,在天然草原自然光下采集典型牧草高光谱数据,对数据进行预处理,利用A -Clenet5法进行特征挖掘和数据分类。研究结果表明,该方法对草种识别精度可达到92.18%,满足基于高光谱的草原草种的识别要求,为草原退化高精度遥感 监测提供了可能。

Identification and classification of desert grassland species based on A-Clenet 5

Grassland degradation has become a maj or ecological problem facing human beings.One of signs is the change of grasslan d vegetation community structure.The hyperspectral recognition and classificatio n of grassland grasses is the basis and premise of remote sensing for large-are a and high-precision grassland degradation monitoring and treatment.Due to the randomness of the distribution of natural grassland species and the complexity o f the remote sensing image cloud system,the problem of low recognition accuracy of grass species has not been satisfactorily resolved.This study proposes the A -Clent5method for the first time based on deep learning,collects typical pastu re hyperspectral data under natural light in the natural grassland.Then the data is preprocessed,and the feature mining and data classification are performed by the A-Cletet5method.The research results show that the recognition accuracy o f this method is 92.18%,which realizes the identification of grassland typical g rass species based on hyperspectral,which provides the possibility of high-prec ision remote sensing monitoring of grassland degradation.