基于局部兴奋全局抑制振荡网络的图像分割

IMAGE SEGMENTATION BASED ON LOCALLY EXCITATORY GLOBALLY INHIBITORY OSCILLATOR NETWORK

将局部兴奋全局抑制振荡网络（ＬｏｃａｌｙＥｘｃｉｔａｔｏｒｙＧｌｏｂａｌｙＩｎｈｉｂｉｔｏｒｙＯｓｃｉｌａｔｏｒＮｅｔ－ｗｏｒｋ，简称ＬＥＧＩＯＮ）应用于图像分割．将侧电势引入振荡子的动态行为，有效地克服了噪声对主要区域的影响．用含噪声的二值图模拟了ＬＥＧＩＯＮ的时间演化．根据大脑区域分割特征聚合原理的相近性、相似性和连通性原则，设计了分割灰度图的网络连接权．用ＨＳＩ色空间设计了突出Ｈ分量的连接权，得到了比基于ＲＧＢ色空间分割更为满意的分割效果．

An image segmentation scheme was presented on the basis of locally excitatory, globally inhibitory oscillator network (LEGION). Based on the concept of the lateral potential, a solution to remove noisy regions in an image was proposed for LEGION. The temporal evolution of every stimulated oscilltor and network properties were illustrated by computer simulation. The dynamic connection weight addressing the grouping principles of proximity, similarity and connectedness was setup. Using HSI space, the dynamic connection weight emphasizing the Hue information was established, and better experimental results than that based on RGB space were obtained.