基于格式塔心理学原理的几何活动轮廓模型  

A Geometric Active Contour Model Based on the Laws of Gestalt Psychology

作　　者：周昌雄[1] 曹丰文[1] 崔鸣[1] 颜廷秦[1] 张尤赛[2] 

机构地区：[1]苏州市职业大学电子信息工程系,苏州215104 [2]江苏科技大学电子信息学院,镇江212003

出　　处：《中国图象图形学报》2008年第5期924-929,共6页Journal of Image and Graphics

基　　金：江苏省社会发展项目(2005DX015J)

摘　　要：基于格式塔心理学原理提出了一种几何活动轮廓模型,并将其应用于图像分割。当轮廓曲线远离目标边界时,应用格式塔心理学目标-背景原则,其能量函数主要由区域间差异性组成;当轮廓曲线位于目标边界附近时,应用格式塔心理学接近性原则,其能量函数主要由区域内一致性组成。该模型符合知觉特性,是几何活动轮廓模型的一般形式,且融合图像区域信息和边界信息。通过侧脑室和肿瘤医学图像分割实验,其结果表明,该模型对模糊边界图像的自动分割具有一定的普适性,能达到满意的分割效果。将该模型应用到多目标的免疫细胞图像分割中,能一次性完成将细胞质从细胞核和体液两种不同背景中分割出来的任务。Based on the laws of Gestalt psychology, a geometric active contour model for image segmentation is proposed in this paper. The energy function is mainly made of inter-region dissimilar properties based on object-background principle when contour curve is far from the object edges. The energy function is mainly made of intra-region similar properties based on proximity principle when contour curve is situated near the object edge. The developed model, which incorporates the information of foreground and background regions, produces perceptual characteristics and is a general form of the geometric active contour model. Experiment on Lateral Ventricle and tumor images show that in general the model can deal with automatic segmentation of weak edge image efficiently. Furthermore, the experiment on immune celt shows that cytoplasm of immune cell is segmented out from two background regions of cell nucleolus and body fluid with the method.

关 键 词：格式塔心理学原理 几何活动轮廓模型 图像分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]