基于区域竞争的水平集快速图像分割算法

Fast image segmentation algorithm based on region competition with level set

从曲线演化的角度提出一种基于贝叶斯(Bayesian)区域统计和区域竞争的自适应变分图像分割模型,该模型使用水平集描述曲线和区域,得到基于Bayesian区域统计信息的能量函数,利用区域竞争曲线演化模型推导出一种快速曲线演化偏微分方程,实现了图像分割。该方法可以同时提取出多类目标,算法具有快速、分割精度高的特点,且易于综合纹理、形状等多种信息对模型进行扩充。此外,能量函数和曲线演化方程是相对独立的,对于不同类型的图像可选用不同的概率模型。实验表明,该方法是一种快速、有效、新颖的图像分割方法。

An adaptive variational image segmentation model based on Bayesian and region competition was presented. Level set method was used to describe the plane curves and partitioned regions, and the energy function was obtained based on Bayesian region statistical information. Then, a new fast partial different equation for curve evolution was deduced to implement unambiguous image segmentation by region competition. The model can extract multi-class objects simultaneously with fast evolving speed and high segmentation precision. Also, it is easy to integrate other image information such as texture and shape into this model. Besides, the energy function and curve evolution equations are independent so that we can choose different probability functions to describe various types of images. The experimental results show that it is a fast, effective and novel image segmentation algorithm.