一种融合双聚类算法的高分辨率遥感影像分割方法

A High Resolution Remote Sensing Image Segmentation Method with Dual Clustering Algorithm

针对高分辨率遥感影像分割易受噪声干扰问题,提出一种简单线性迭代聚类与区域动态约束聚类相融合的双聚类算法。该法首先利用简单线性迭代聚类产生超像元,然后运用区域动态约束聚类算法对超像元进行聚类合并,结合聚类生成的方差和、局部方差和局部方差变化率,确定合理分割数,最终完成影像分割。实验表明,本文提出的影像分割方法抗噪声干扰能力较强,且分割结果精度较高,未出现欠分割和过分割现象,Kappa系数高达0.8312,OCE值低至0.4185,属于一种稳健的遥感影像分割方法。

For the problem that the segmentation of high resolution remote sensing image is liable to noise interference,a dual clustering method based on simple linear iterative clustering and regionalization with dynamically constrained agglomerative clustering and partitioning is proposed.This method firstly uses a simple linear iterative clustering to produce super pixels,then uses the clustering and partitioning algorithm to cluster merging for the super pixel,combined with sum of squared deviations,local variance,and rate of LV change,the reasonable segmentation number is determined,and the image segmentation is finally completed.Experiments showed that the proposed image segmentation method has strong anti-noise ability and high accuracy of segmentation results,and there is no under segmentation or over segmentation.The Kappa coefficient is as high as 0.8312 and OCE value as low as 0.4185,it is a robust segmentation method for remote sensing image.