一种融合颜色和空间信息的区域生长算法

Region Growing Using Color and Spatial Information

该文提出了一种基于图像颜色和局部空间信息的种子区域生长算法,并用于彩色图像分割。该算法首先根据相对欧式距离使用均值聚类算法对图像进行颜色量化,形成图像的初始分割结果,然后通过计算局部颜色散度,进行分级区域合并,最后,利用形态学相关算法对分割区域的边缘进行平滑。实验表明,该算法能得到与人类视觉判断相一致的有意义区域的分割。

In this paper, the seed region growing algorithm based on the color of the image and spatial information is put forward and it goes for the color image division.First of all, with Euclidean distance the algorithm uses the relative means clustering algorithm to quantify the color image, which results in the initial image segmentation. Then by calculating the divergence of partial color, the grading region is to be combined.Finally, smooth the edge of the region of the segmentation with the algorithm related to morphology.The experiments show that with the algorithm, the meaningful segmentation region corresponded with human's vision judgment can be achieved.