目前半监督高光谱影像分类方法中,针对未考虑标记样本邻域信息、迭代训练中样本量的研究较少,因此提出了一种基于空间邻域信息和多分类器集成的半监督高光谱影像分类方法;并分别基于Salinas Valley AVIRIS和Pavia ROSIS高光谱影像数据集对该分类方法进行了实验分析。结果表明:(1)初始模型训练时通过对标记样本进行邻域生长,保证初始模型训练的可靠性;(2)通过SVM和MLC的预测结果综合评估非标记样本,可实现精准标记;(3)对于模型训练样本的选择,通过抽取部分样本对分类模型进行多次投票试验,取票数多的作为最终分类结果,保证了分类结果的准确性。该方法能很好地改善较少样本情况下图像分类模型的泛化能力,提高分类精度。
Cavalli Rosa Maria.Capability of Remote Sensing Images to Distinguish the Urban Surface Materials: A Case Study of Venice City[J].Remote Sensing,2021.
Elcoroaristizabal S.,Amigo J.M..Near infrared hyperspectral imaging as a tool for quantifying atmospheric carbonaceous aerosol[J].Microchemical Journal,2020.
Kun Tan,Jun Hu,Jun Li,Peijun Du.A novel semi-supervised hyperspectral image classification approach based on spatial neighborhood information and classifier combination[J].ISPRS Journal of Photogrammetry and Remote Sensing,2015.
Wang, L.,Hao, S.,Wang, Y.,Lin, Y.,Wang, Q..Spatial–Spectral Information-Based Semisupervised Classification Algorithm for Hyperspectral Imagery[J].IEEE journal of selected topics in applied earth observations and remote sensing,2014.
Inmaculada Dopido,Jun Li 0009,Prashanth Reddy Marpu,Antonio Plaza,José M. Bioucas-Dias,Jon Atli Benediktsson.Semisupervised Self-Learning for Hyperspectral Image Classification.[J].IEEE Trans. Geoscience and Remote Sensing,2013.
Lijun Ding,Ardeshir Goshtasby.On the Canny edge detector[J].Pattern Recognition,2001.
Vladimir Vapnik.An overview of statistical learning theory.[J].IEEE Trans. Neural Networks,1999(5).
Behzad M. Shahshahani,David A. Landgrebe.The effect of unlabeled samples in reducing the small sample size problem and mitigating the Hughes phenomenon.[J].IEEE Trans. Geoscience and Remote Sensing,1994(5).