摘要
同步器齿环具有不规则和一定的统计自相似性的特征,可用分形方法中的分形维数对齿环图像进行描述,并将维数作为测量特征实现齿环缺陷识别。差分盒维法计算齿环图像维数时,结果会受到背景和摆放角度的影响,导致正常与缺耳齿环的维数结果存在交叉现象。针对这个问题,在研究和分析差分盒维数法的基础上提出一种改进差分盒维法,即利用概率和图像旋转消除图像背景和被测物摆放角度的影响,提高维数计算精度。通过改进差分盒维法对正常与缺耳齿环的分形维数计算与分析,可消除图像背景和摆放角度对维数产生的影响,实现了用齿环分形维数作为识别特征量对缺陷齿环的识别。
synchronizer rings（gear ring）present irregular shape and certain self-similar characteristics,so we can use fractal dimension to describe the gear ring images and achieve defect recognition as a measurement characteristics.When calculating the dimension of gear ring images with differential box-counting method,the results are effected by image background and display angle,leading to the cross phenomenon on dimensions of normal and missing ear gear rings.To solve this problem,this paper proposes a rotational differential box-counting method based on research and analysis of differential box-counting method.This method utilizes probabilistic and image rotation methods to eliminate the effects caused by image background and display angle,and improve the dimension accuracy.Fractal dimension of normal and missing ear gear rings are estimated by the rotational differential box-counting method.The results show that the effects of background and angle are eliminated and fractal dimension can be characterized as a measurement of normal and defective gear ring identification.
出处
《湖北工业大学学报》
2016年第2期5-8,共4页
Journal of Hubei University of Technology
基金
国家自然科学基金(51275158)
关键词
同步器齿环
改进差分盒维法
分形维数
缺陷识别
synchronizer gear ring
rotational differential box-counting method
fractal dimension
defect detection
