基于模糊聚类和形态学的轮胎断面特征提取

Feature extraction of tire section based on fuzzy clustering and morphology

针对轮胎断面图像中胶料之间灰度相差小,边界模糊性强,传统的边缘提取方法难以实现断面特征提取的问题,提出了1种将多层次模糊增强、模糊C均值聚类以及形态学等方法有机结合,用于轮胎断面特征提取的算法.该算法先将钢丝区域去除并对图像整体进行灰度变换,然后通过多层次模糊增强,增加胶料之间的对比度,再运用模糊C均值聚类分割出各胶料区域,并对各区域进行形态学处理并提取边界,最后将各区域边界叠加,加入钢丝区域,得到轮胎断面的特征.对不同光照条件下的轮胎断面图像进行试验.结果表明:该算法可以解决轮胎断面图像中胶料之间灰度差小的问题,并有效提取了轮胎断面中的胶料边界和钢丝特征,且对光照条件具有一定的适应性.

The features of tire section were difficult to be extracted by the traditional edge detection me-thods because of the small difference of gray value between rubbers in the tire section image and the fuzzy boundaries. An algorithm was proposed to extract the features of tire section by combining multi-level fuzzy enhancement, fuzzy C-means clustering and morphology. The regions of steel wires were removed to transform the gray values of image. The multi-level fuzzy enhancement was used to increase the contrast between rubbers. The rubber regions were effectively segmented by fuzzy C-means clustering, and the boundaries were extracted by morphology. The features of tire section were obtained by stacking the boundary of each region and adding steel wires. Experiments of tire section image were conducted under different light conditions. The results show that the algorithm can effectively solve the problem of small gray value difference between rubbers in the image of tire section and extract the features of tire section with good applicability to different light conditions.