Abstract:

Simple Linear Iterative Clustering（SLIC） superpixel algorithm relies on the number of superpixel settings, which is prone to under segmentation or over segmentation, and also a low efficiency. An improved Simple Non-Iterative Clustering（SNIC） superpixel algorithm for segmentation of remote sensing images is proposed in. Firstly, SNIC superpixels are used to obtain the initial segmentation results. Secondly, the dynamic threshold is applied to segment the original image, and then the images are made difference twice so as to correct the SNIC segmentation results. Finally, the segmentation lines that meet certain conditions are the final segmentation results. Experimental results show that the proposed algorithm has obtained good results in terms of precision, recall and running time.

Key words: remote sensing images, image segmentation, Simple Non-Iterative Clustering（SNIC）, dynamic threshold

摘要：

简单线性迭代聚类（Simple Linear Iterative Clustering，SLIC）超像素算法依赖超像素设置数目的大小，容易产生欠分割或者过分割的现象，且运行速度不高。提出一种改进的简单非迭代聚类（Simple Non-Iterative Clustering，SNIC）超像素算法对遥感影像进行分割。采用SNIC超像素获取初始分割结果；利用动态阈值对原始影像进行分割；对影像进行两次作差，从而对SNIC分割结果进行修正；选取满足一定条件的分割线即为最终的分割结果。实验结果表明，该算法在分割精度、召回率和运行时间上都获得了令人满意的结果。

关键词: 遥感影像, 图像分割, 简单非迭代聚类, 动态阈值