基于改进亲和度图的矿石颗粒图像分割研究与实现

Research and Implementation of Ore Particles Image Segmentation Based on Improved Affinity Graph

针对矿石颗粒的图像分割,直接使用已有的图像分割算法难以满足分割要求。为了进一步提升矿石颗粒图像分割算法的有效性,采用超像素与邻域超像素之间的线性表达建立一种新的邻域亲和度图。首先,对原始图像进行过分割得到几组尺度不同的超像素,并提取超像素的颜色和纹理特征;然后,利用超像素的邻域字典来求解改进的邻域亲和度图,用以描述超像素间的相似度关系;最后,引入二部图来表达像素和超像素之间的归属关系,同时利用Tcuts(transfer cuts)算法进行分割。实验结果表明:该算法在碎矿石颗粒图像中分割效果要优于其他现有的算法,对光照变化和噪声有一定的鲁棒性。

For image segmentation of ore particles,it is difficult to meet the requirements of segmentation by directly using existing image segmentation algorithms.To further improve the effectiveness of the ore particles image segmentation algorithm,the novel neighborhood affinity graph was constructed by linear expression between the superpixel and neighborhood superpixel.Firstly,the original image was segmented to obtain several sets of superpixels with different scales,and the color and texture features of the superpixels were extracted.Then,the neighborhood affinity graph was built by using the neighborhood dictionary of superpixels,which described the similarity relationship between superpixels.Finally,a bipartite graph was introduced to express the affiliation relationship between pixels and superpixels,and segmentation is performed using the Tcuts(transfer cuts)algorithm.The experimental results show that the segmentation performance of the proposed method is better than other existing algorithms in ore particles image segmentation,and the algorithm is robust to illumination changes and noise.