基于超像素的高分遥感影像分割算法

High-resolution remote sensing image segmentation algorithm based on superpixel

针对高分遥感影像中存在地物数目多,特征信息复杂导致分割边缘不清晰、对象细节丢失等问题,提出一种改进的超像素分割和多特征结合的遥感影像分割合并算法。在对图像进行分割前的预处理阶段,使用超像素分割技术得到初始分割图像;区域合并过程中,基于对象间的异质性和对象内部的同质性,结合光谱、纹理和形状特征,对对象进行合并;通过调整全局分割参数来调整合并尺度,得到最终的影像分割结果。实验结果表明,所提方法能得到较好的影像分割效果。

For the high score remote sensing image,there are many features,and the feature information is complicated,resulting in unclear edge and loss of object details.An improved superpixel segmentation and multi-feature combination remote sen-sing image segmentation and merging algorithm was proposed.In the pre-processing stage,the initial segmentation image was obtained using the superpixel segmentation technique.In the process of region merging,objects were merged based on the hete-rogeneity between objects and the homogeneity within the object,combined with spectral,texture and shape features.The merged scale was adjusted by adjusting the global segmentation parameters to obtain the final image segmentation result.Experimental results show that the proposed method can get better image segmentation effects.