多分辨分形理论在高分辨力遥感图像分割中的应用

Multifractional image segmentation in high resolution remote sensing image

针对遥感图像的高分辨力特性,提出了基于多分辨分形的遥感图像分割方法。改进了多分辨结构的建立方法,提出了基于Laplacian金字塔的多分辨结构的分形分割法。实验表明,改进的多分辨分形图像分割方法在减少信息冗余度的同时,提高了数据利用率,可有效地对高分辨力遥感图像进行分割分类。

According to the high resolution of remote sensing images,an image segmentation based on multifraction is presented. A model of multiscale structure is modified, and the structure based on Laplacian pyramid is introduced in fractional image segmentation. The experiment results on remote sensing image segmentation show that the improved method reduces the in- formation redundancy, increases the data utilization, and effectively segments and classifies high resolution remote sensing image.