基于JSEG改进算法的高分辨率遥感影像分割

Segmentation of high resolution remote sensing images based on improved JSEG algorithm

高分辨率遥感图像中细节信息丰富、地物几何结构明显,对JSEG算法进行了改进,使其对高分辨率遥感图像分割可以取得更合理的结果。算法使用增量式的生长方式完成初始分割,并综合使用颜色和形状信息对过分割区域进行合并。实验表明,改进后的算法符合高分辨率遥感图像的特点,可以得到更好的效果。

More reasonable segmentations on high resolution remote sensing images are generated from an improved JESG algorithm in terms of the characteristics of rich detail information and clear geometric structure. Our algorithm employs an incremental growing approach to accomplish initial segmentation. And an integrated criterion including the color and shape information is used to merge the over-segmented regions. Experimental results demonstrate that the improved algorithm gains better performance because of the complete use of the metrics of high resolution remote sensing images.