基于NL-Means的均值平移图像分割算法

Mean Shift Image Segmentation Based On NL-Means

针对均值平移图象分割算法中,在密度中心点选择时的不足,本文采用一种新的寻找密度中心点的方法,同时,为了克服传统基于特征空间分析的图像分割方法对像素点空间关系考虑不够充分的缺陷,通过Non-local means算法,在距离公式中引入特征权参数,从而优化聚类效果。对图象分割结果分析表明了这种方法的有效性。

Aim at the shortage of density central point choice in Mean shift image segmentation, this article uses a new way of searching for the density central point, in the meantime, in order to overcome the blemish of the space relation to the pixel of tradition feature space analytical ,during Non-local means algorithm ,using characteristic weight parameter in the distance formula ,thus make the clustering effect better the experiment is shown thai the proposed algorithm can be valid.