结合区域合并和水平集的图像分割方法

Image segmentation method combining regional merger and level set

提出了一种新的图像分割方法.首先利用均值漂移方法对图像进行分割处理,在此基础上进行区域合并得到图像粗分割,然后利用粗分割的结果构造初始水平集函数,利用水平集方法获得图像的最终分割结果.并且针对水平集方法提出了一种新的边缘检测算子,重新定义边缘停止函数,使得对受到噪声影响图像的分割具有鲁棒性.实验结果表明,本方法能有效地提取图像目标,在图像分割质量上具有一定优势,具有较强的实用性.

A novel image segmentation is proposed. Firstly, the mean shift algorithm is applied to the image, the region merge method,which takes the result of the mean shift,is used to capture the coarse segmentation, and then the coarse segmentation result is used to construct the initial level set function. Finally, the level set method is used to obtain the final image segmentation results, this paper proposes a new edge detection operator to redefine a new edge stopping function. It is more robust to noisy image segmentation. Experimental results show that this method can extract mage target effectively, has some advantages in the quality of image segmentation.