Abstract: In the process of using traditional Mumford-Shah model in image segmentation，there are always two elliptical equations in each evolution which need to be solved，the solving process seriously influences the speed of algorithm.In order to overcome the shortcoming，a slicing constant function is used to substitute the output image in M-S model，so the calculation is effectively simplified.The experiments show that the algorithm proposed has the same results as the traditional method，but excels in speed.

Key words: Mumford-Shah model, level set, slicing constant function

摘要： 针对传统的Mumford-Shah（M-S）模型在图像分割的每一次迭代中都需要另外求解两个椭圆型方程，从而使得算法的执行效率大大降低的缺点，利用分片常函数代替M-S模型中的待优化输出图像，简化了曲线演化的过程，避免了额外计算，有效地提高了算法的执行速度。实验表明，该方法在图像分割效果方面与传统方法没有明显差别，但是程序的运行速度方面却远远优于传统方法。

关键词: Mumford-Shah模型, 水平集, 分片常函数