基于知识的多层Mumford-Shah向量值图像分割模型

Knowledge Based Hierarchical Mumford-Shah Model for Vector-valued Image Segmentation

针对多层Mumford-Shah模型不能正确分割对比度小且部分被遮挡的复杂医学图像问题,将目标形状先验知识窄带水平集统计形状模型集成到多层Mumford-Shah模型,提出了基于目标形状先验知识的多层Mumford-Shah向量值图像分割模型和求解该图像分割模型泛函最小值的水平集逐层迭代算法.实验结果表明,该方法能够有效分割对比度小且部分被血管遮挡的早期青光眼病人视乳头图像.

To segment a given vector-valued image such as color images, and to handle important image features such as very low contrast and obscured part, a novel knowledge based hierarchical Mumford-Shah functional model is addressed by integrating a statistical shape prior model based on narrow band level set with hierarchical Mumford-Shah model. At the same time, an iterative tier-by-tier algorithm based on techniques of level set is proposed to minimize the new functional. This novel model could recognise an object whose shape is similar to the prior one. The experimental results show that the technique is effective and practicable by applying it to the segmentation of the optic disk obscured by blood vessels in color optic nerve head images of early stage glaucoma patients.