为了有效地解决血管遮挡、噪声污染、光照不均、对比度小以及个体间差异大等视乳头图像分割中固有的难题,提出采用基于图论的多相分段常数水平集Mumford-Shah图像分割模型及其相应的图分割最优化方法。实验表明,该方法能够比经典的多相分段常数水平集Mumford-Shah模型更快更精准地分割青光眼病人视乳头图像中的视杯和视盘形态。
In order to solve the existing problems caused by blood vessels obscuration,noise pollution,illumination odds,low contrast,and inter-differences when one segments the optic cup and disk in optic nerve head images,this paper proposed to use graph min-cut method for multiphase Mumford-Shah model.The tests demonstrate that the proposed method is more efficient and more accurate than PCLSM for multiphase Mumford-Shah model.