一种Mumford-Shah模型的图像分割改进算法

提出了一种新的基于Mumford-Shah模型的图像分割改进算法，并通过对飞机图像轮廓分割进行效果验证。首先，内部惩罚项的设置，彻底消除了计算量巨大的更新水平集函数的步骤。第二，可使用较大的时间步长，获得较快的曲线演化速度。第三，水平集函数可初始化为更容易构造和使用的一般函数，初始轮廓线可以构造为任意形状。最后，选择因子的设置既继承了克服轮廓线“冒顶”的性能，又保证了曲线未达轮廓边界时的演化速度。

An Improved Algorithm Based on Mumford-Shah Model for Image Segmentation

This paper describes an improved algorithm based on Mumford-Shah model for image segmentation and demonstrates it by segmenting the airplane contour.The new level set method has four main advantages over the traditional.First,we use an internal energy term that penalizes the deviation of the level set function from a signed distance function and therefore the need of the costly re-initialization procedure is completely eliminated.Second,a significantly larger time step can be used for numerically solving the evolution partial differential equation,and therefore speeds up the curve evolution.Third,the level set function can be initialized with general functions that are more efficient to construct and easier to use in practice.Finally,we set the choice coefficient to prevent some contours from exceeding the object boundary whereas others still in evolution and speed up the evolution when the contour don't arrive at the object boundary.