MeanShift算法在图像分割中的应用  

Application of MeanShift Algorithm in Image Segmentation

作　　者：韩俊 王保云 Han Jun;Wang Baoyun(Yunnan Normal University,School of Information Science and Technology,Kunming 650500;Yunnan Normal University,School of Mathematics,Kunming 650500)

机构地区：[1]云南师范大学信息学院,昆明650500 [2]云南师范大学数学学院,昆明650500

出　　处：《现代计算机》2021年第33期71-76,共6页Modern Computer

基　　金：国家自然科学基金:基于深度迁移学习的遥感影像中泥石流孕灾沟谷识别——以云南省为例(61966040)。

摘　　要：对MeanShift算法的基本原理、改进方法进行了概括与总结,并对其在图像分割中的应用进行了详细的分析。文中选取目标和背景均较为简单、目标区域较为明显但背景复杂、多目标实例且背景复杂三类图像进行算法性能验证,并与常见的图像分割算法进行实验对比。实验结果表明,MeanShift算法对三类图像的分割效果都比较好,能够胜任大多数场景的图像分割任务。The basic principle and improvement method of MeanShift algorithm are summarized and summarized, and its application in image segmentation is analyzed in detail. In this paper, three types of images, which are relatively simple target and background,more obvious target area but complex background, multiple target instances and complex background, are selected for algorithm performance verification, and are compared with common image segmentation algorithms. The experimental results show that the MeanShift algorithm has better segmentation effects for the three types of images, and can be competent for image segmentation tasks in most scenes.

关 键 词：图像分割 聚类 MEANSHIFT 核函数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]