区域进化自适应高精度区域增长图像分割算法

为克服经典区域增长算法中门限选择困难、分割稳定性不高与串行处理速度慢的不足,提出基于区域进化的自适应高精度区域增长图像分割算法。在图像预处理过程中,首先通过各向异性滤波算法对切片进行滤波,达到去除图像噪声同时避免对边界区域的模糊;然后引入了新的区域能量表示模型,并给出了迭代进化形式,在区域增长过程中,逐渐增加区域增长的门限,通过对能量函数的动态优化来逼近最佳分割结果;最后利用主动轮廓模型进行精度分割,得到精确而比较光滑的分割目标轮廓。对比实验表明提出的方法是合理有效的。

Adaptive and high accurate region growing image segmentation method based on region evolution

To overcome the difficulty of manual threshold selection and slow speed of conventional region growing image segmentation algorithm.An adaptive and high accurate region evolution image segmentation algorithm is proposed.Firstly the anisotropic diffusion filter is used to preserve the edge information.Then the novel region energy model and the corresponding iterative evolution steps are given to obtain coarse contour.In region growing processing step,the threshold is increased gradually,so the optimal segmentation results are obtained by dynamic optimizing the energy model.Finally in order to achieve accurate segmentation results,the active contour model is exploited to further segment to get accurate boundary.Comparative experimental results show that this algorithm performs better obviously in automatic optimal threshold selection,processing speed and segmentation veracity than conventional region growing algorithm.