基于Mean Shift和随机游走的图像分割算法

提出了一种改进的随机游走算法。首先,采用Mean Shift算法对图像进行预处理,将图像划分成一些同质区域,用同质区域作为节点进行随机游走,在降低节点数的同时也抑制了噪声对分割的影响;其次,利用马氏距离定义区域之间的权值;对种子点进行了改进,增加了辅助种子点,利用辅助种子点和用户标记的种子点进行随机游走,实现同质区域的合并,实现图像的最终分割。实验结果表明,该算法提高了图像分割的精度。

Image Segmentation Algorithm Based on Mean Shift and Random Walk

An improved random walk algorithm was proposed herein.First,Mean Shift algorithm was adopted to preprocess the image,which was partitioned into a series of homogeneous areas,so that the homogeneous areas were taken as nodes to walk at random,with noise inhibited while reducing the number of nodes.Second,PMD was used to define the weight between regions.Thirdly,seeds were improved to have added the auxiliary seeds,and the auxiliary and signed seeds were used to walk random,with region merging realized.The final image segmentation was reached.Experimental results expatiates that the proposed method highlights the segmentation accuracy.