一种基于统计学习理论的最小生成树图像分割准则

A Image Segmentation Method Based on Statistics Learning Theory and Minimum Spanning Tree

根据基于区域增长的面向对象图像分割的本质特点,将统计学习理论与最小生成树算法相结合,提出了一种基于统计学习理论的最小生成树图像分割准则。将该图像分割准则应用于多种遥感影像数据进行分割实验,其结果表明基于统计学习理论的最小生成树图像分割准则能通过简便的参数设置,即可以较好地实现不同尺度目标的图像分割,同时又能对纹理区域进行有效分割,能获得良好的区域边界和较好的抗噪声性能,并在海岸带大比例尺无人机正射影像的图像分割实践中得到了较好验证。

According to the essential feature of object-oriented image segmentation method,this paper explores a minimum span tree（MST）based image segmentation method.We define an edge weight based optimal criterion（merging predicate）which based on statistical learning theory（SLT）,a scale control parameter is used to control the segmentation scale.Experiments based on the high resolution UAV images show that the proposed merging predicate can keep the integrity of the objects and do well on preventing over segmentation.It also proves its efficiency in segmenting the rich texture images while can get good boundary of the object.