Abstract: An image segmentation method which combines edgeflow and fuzzy region clustering is put forward. An image is segmented by high precision of gray edgeflow to obtain the vector directions and edge energies of the edgeflow. The vector directions and edge energies are then used to get the initial segmented image through curve evolution. The small areas generated by the initial segmentation are clustered by using the joint distribution of color space and image space. The results of the clustering are used to correct the initial segmented image to remove over-segmentation. The experimental results show that this method can obtain good effect in core image segmentation.

Key words: edgeflow, clustering, image segmentation, curve evolution

摘要： 提出一种融合边缘流和模糊区域聚类的图像分割方法。选用高精度的灰度边缘流对图像进行分割，得到边缘流的矢量方向和边缘能量，通过曲线演化得到初始分割图像；在初始分割产生的小区域上，综合小区域的色彩空间特征和图像空间特征，进行了模糊区域聚类；将聚类结果用于修正初始分割图像，去除过分割。实验表明，在对岩心图像的分割中，该方法能取得良好的效果。

关键词: 边缘流, 聚类, 图像分割, 曲线演化