红树林生态系统不仅是重要的固碳载体，还具有维护滨海湿地生物多样性、促淤固滩、净化水体等生态功能。珠海市淇澳岛是珠三角为数不多的红树林分布区，对该地红树林树种组成进行调查和识别对湿地生态系统的保护和利用具有现实意义。本研究基于珠海一号高光谱遥感数据，开展淇澳岛红树林优势种识别研究。研究设计基于像素和面向对象两种分类方案，基于像素的方法挖掘影像光谱信息，采用波谱角填图算法进行优势种识别；面向对象的方法结合光谱和纹理信息训练支持向量机进行分类。结果表明，面向对象的分类效果更优，总体精度为80.95%，Kappa系数为0.72，体现了高光谱数据结合面向对象方法的种间分类潜力，也为红树林资源调查提供案例参考，利于红树林生态系统的保护与开发利用。

The mangrove ecosystem is not only an important carbon sink, but also plays crucial roles in maintaining coastal biodiversity, promoting sediment accumulation, and purifying water bodies. Qi'ao Island in Zhuhai City is one of the few mangrove distribution areas in the Pearl River Delta. Investigating and identifying the tree species composition of mangroves in this area holds practical significance for the conservation and utilization of wetland ecosystems. This study is based on OHS hyperspectral remote sensing data, conducting research on the identification of dominant mangrove species on Qi'ao Island. The research design includes two classification schemes: pixel-based and object-oriented methods. The pixel-based approach explores image spectral information and employs the Spectral Angle Mapper algorithm for species identification, while the object-oriented method combines spectral and texture information to train Support Vector Machines for classification. Results indicate that the object-oriented classification method outperforms, with an overall accuracy of 80.95% and a Kappa coefficient of 0.72, demonstrating the potential of combining hyperspectral data with object-oriented methods for inter-species classification. This study also provides a case reference for mangrove resource surveys, facilitating the conservation and sustainable development of mangrove ecosystems.