组合分类器及其在高光谱影像分类中的应用

Combined Classifiers and Its Application in Hyperspectral Classification

多分类器组合问题一直是模式识别和机器学习领域的研究重点。由于各分类器的分类结果有着互补信息,因而一般来说,组合分类方法总有着较单个分类器精度更高的分类结果。将组合分类方法引入到高光谱影像的分类过程中,并针对该类影像的特点,提出了混合组合策略。实验表明,将组合分类器用于高光谱影像的分类,可以得到较为理想的分类结果。

Research on combined classifiers has been very important in pattern recognition and machine learning. It is because that these classification results are complementary which generally leads to a better classification performance in terms of combined classifiers. In this paper the author introduced the combined classifiers method into the hyperspectral classification and presented the hybrid combination rule. Experiment demonstrated that combined classifiers were more powerful and finally more accurate classification result achieved.