基于核特征距离的局部活动轮廓模型

Active contour model by kernel-induced distance fitting energy

提出了一种基于核特征距离局部活动轮廓分割模型。在模型中使用核特征距离来构造局部拟合能量,从而可以获取精确的局部图像特征,可以分割存在灰度不均匀的图像。并通过引入水平集规范项以避免水平集演化的重新初始化,提高了分割的效率。实验结果表明,本模型可以很好地克服灰度不均匀性,同时在分割精度上有了较大的提升,特别是分割速度比LBF模型快1.3～1.5倍。

This paper proposed a new active contour model based on kernel-induced distance for image segmentation.It introduced a local fitting energy with a kernel-induced distance,which enabled the extraction of accurate local image information.Therefore,this model could be used to segment images with intensity inhomogeneity.In addition,it used the level set regularizing term to avoid expensive reinitialiation of the evolving level set and improved the segmentation speed.Experimental results demonstrate that the proposed algorithm can overcome intensity inhomogeneity,and get the more accurate segmentation result,especially there is about 1.3 to 1.5 times faster than the LBF model.