论文部分内容阅读
针对多尺度分割技术中的最优尺度选择问题,提出一种基于分形网络演化方法和改进模糊聚类遥感影像分割的方法。该方法利用分形网络演化方法对原始影像进行小尺度分割,并利用粒子群算法的全局搜索能力,从预分割的小尺度对象中确定最优初始聚类中心,在对小尺度对象聚类合并时,建立具有对象空间信息和对象间相关信息的目标函数,最终得到适应不同尺度地物的分割结果,降低了多尺度分割方法对尺度参数的过度依赖。实验结果表明,该方法可获得高质量的遥感影像分割结果。
Aiming at the problem of optimal scale selection in multi-scale segmentation, a method based on fractal network evolution and improved fuzzy clustering remote sensing image segmentation is proposed. This method uses the fractal network evolution method to segment the original image into small scales and uses the global search ability of the particle swarm optimization algorithm to determine the optimal initial cluster centers from the pre-segmented small-scale objects. When clustering the small-scale objects , An objective function with object space information and related information between objects is established. Finally, the segmentation results of features adapted to different scales are obtained, which reduces the excessive dependence of the multi-scale segmentation method on the scale parameters. Experimental results show that this method can obtain high-quality remote sensing image segmentation results.