基于支持向量机的轮胎标识点形状识别  

The Shape Recognition of Tyre Marking Points Based on Support Vector Machine

作　　者：王勇[1] 郭慧[1] 

机构地区：[1]华东理工大学机械与动力工程学院,上海200237

出　　处：《图学学报》2014年第2期274-279,共6页Journal of Graphics

摘　　要：轮胎标识点在轮胎分类中发挥着重要的作用,为了解决人工识别标识点工作强度大、误差较高等问题,提出了一种基于支持向量机的轮胎标识点形状识别算法。首先利用中值滤波对标识点图像进行降噪处理,根据标识点图像的灰度直方图对图像进行分割以去除背景,并利用Canny边缘检测算子提取标识点边缘轮廓。然后,提取标识点轮廓的傅立叶描述子系数,最后利用支持向量机对标识点的傅立叶描述子系数进行分类以识别标识点的形状。实验结果表明,该算法能准确地识别出轮胎标识点的形状,提高了标识点识别效率。The tyre marking points play an important role in the tyre classification. However, manual recognition is laborious and is easily influenced by physiological factors. A method of shape recognition of tyre marking points based on Support Vector Machine(SVM) is proposed in this paper. Firstly, the image of tyre marking points will be denoised by median filter and segmented based on the gray level histogram. Then the edge of tyre marking points will be detected by Canny algorithm, and the detected edge will be represented by Fourier transform. Lastly, those Fourier coefficients will be used as input to SVM classifier for distinguishing shapes of tyre marking points. The experimental results show that the proposed method can achieve high correct shape recognition rate and improve the efficiency of recognition of tyre marking points.

关 键 词：支持向量机 标识点 傅立叶描述子 形状识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]