一种基于BP网络和决策融合的高光谱遥感图像分类方法

A hyperspectral remote sensing image classification based on BP neural network and decision fusion

BP神经网络在用于高光谱遥感图像分类时,其初始权值的选取对分类结果有很大影响.针对这种情况,提出了一种将BP神经网络与决策融合理论相结合的高光谱遥感图像分类方法,该方法将多个结构相同、初始权值不同的BP神经网络的分类结果进行融合,最后把融合结果作为原图像的最终分类结果,以实际的高光谱遥影像为例,说明该方法能够有效地提高遥感影像的分类精度.

The initial weights of BP neural network have considerable effect on the classification accuracy when BP neural network is used in the classification of hyperspectral images. This paper presents a classification method that combines BP neural networks with decision fusion theory. This method fuses together several classification results obtained from BP neural network for the same structures but with various initial weights. Finally, the fused results is employed as the terminal classification results of primary images. Experimental results on real hyperspectral remote-sensing images show that this method is effective to improve the classification accuracy.