GrabCut彩色图像分割算法的研究

Research of Grabcut Color Image Segmentation Algorithm

GrabCut算法是一种高效的前景背景分割算法。传统分割算法应用纹理或边界信息中的一种进行分割,而GrabCut算法综合利用了纹理及边界两种信息进行图像分割。GrabCut算法是对Graphcuts算法的改进,具有分割精度高,交互式操作少的优点,并将分割从灰度图像推广到了彩色图像领域,是一种比较有前途的交互式分割算法。该算法用在静态图像编辑中的图片背景更换,取得了很好的效果。

GrabCut is an efficient image segmentation algorithm. Both of the information of texture and edge are used in GrabCut, while only one kind of information is used in classic segmentation algorithm. GrabCut is an improvement of Graph cuts, and has high segmentation precision, and needs less interactive operations. Meanwhile, GrabCut is not only used in original grey image, but also used in the field of colorful image. Therefore, GrabCut is a promising interactive segmentation algorithm, and gets good results when being used in background replacement in the static picture editing.