具有多指标柔性能量的Mumford Shah模型图像分割

针对Chan和Vese提出的基于Mumford Shah泛函的水平集图像分割算法，做了两方面的改进：首先,构造了具有柔性的演化曲线内外能量取代C V模型中的刚性能量，减少了C V模型求解时的数值不稳定和过度分割等现象；其次，综合图像的多方面特征，提出多指标集能量项构造方法，提升了C V模型的分割能力和精度。综合两方面的工作，提出带多指标柔性能量的C V模型。新模型能有效处理图像受严重噪音污染、目标内部有灰度起伏等情况。对人工合成图像、医学图像和真实世界图像的分割实验均表明了新模型的良好性能，并且算法收敛速度快、数值稳定。

Mumford-Shah approach for image segmentation with multi-index flexible energy

The Mumford-Shah functionality model is more and more important for image processing.Chan and Vese proposed a level set method based on the simplified Mumford-Shah model for image segmentation.C-V model was improved in two respects:First,new inner and outer energy with flexibility were constructed,instead of traditional rigid energy,which decreased the phenomena of numerical oscillation and deeper segmentation;Second,inner and outer energy based on multi-image information were constructed to improve the segmentation ability and precision of C-V model.Finally,a Mumford-Shah based approach for image segmentation with multi-index flexible energy was proposed via above two points.New model can deal with the image with high noise pollution and inner gray variety.The experiments on synthesized images,magnetic resonance image(MRI) and real world images show the capability of this method,and it is faster convergent and robust.