一种快速高分辨率遥感影像分割算法

Fast Segmentation Algorithm of High Resolution Remote Sensing Image

高分辨率遥感影像(如IKONOS影像)海量数据、复杂细节的特点决定了高分辨率遥感影像分割的技术难点,提出了基于同质性梯度特征、分水岭算法和最小代价合并的快速分割方法。首先对于原始图像进行同质梯度计算得到同质梯度图像;其次利用一种高效的分水岭变换获得初始分割图像;最后给出一种改进的区域合并算法来优化初始分割区域。应用于IKONOS影像的实验证明与其他的分割算法相比,采用所提出的分割方法能快速、准确地获得高分辨率遥感图像的分割结果。

The characters such as large size and complex details of high resolution remote sensing image like IKONOS image determines the technological difficult point of such image segmentation. In this study, a fast and accurate segmentation approach was proposed based on homogeneity gradient character, watershed transform and min-cost region merging. First, a homogeneity gradient image was produced from initial image. Then, an efficient watershed transform was employed to gain the initial segments. Finally, an improved region merging approach was proposed to merge the initial segments and the final segment was obtained. Experiments show, compared with other segment approach, the proposed one is a bit faster and a bit more accurate when applied to the IKONOS image.