双模态脉冲耦合神经网络高分辨率光学卫星影像分割

High Spatial Resolution Optical Satellite Image Segmentation Based on Double-Mode PCNN

针对应用PCNN分割高空间分辨率光学卫星影像存在的问题,提出一种双模态PCNN算法。利用北京地区QuickBird影像进行实验,结果表明,该算法能够弱化影像目标内部灰度变化信息对结果的影响,并能提取影像目标几何结构特征信息,为高空间分辨率光学卫星影像分割提供了一种新方法。

A double-mode PCNN was proposed to segment the high spatial resolution optical satellite image. Experiments were carried out on a subset of QuickBird image covering Beijing. Results showed that this algorithm could alleviate the negative impact on segmentation due to spectral variations of the target, and well keep geometrical information of the target. It provides a promising tool for high spatial resolution optical satellite image segmentation.