轮胎纹理提取与缺陷检测方法

Tire Texture Extraction and Defect Detection Method

为提高检测的智能化水平,以常见的轮胎纹理缺陷为对象,给出2类基于模板的轮胎缺陷检测方法。第1类采用灰度共生矩阵和直方图统计距提取轮胎的纹理特征,第2类采用一维Gabor小波提取轮胎的纹理特征。通过提取并比对模板和待检图像的纹理特征,进行缺陷检测。实验结果表明,第2类检测方法在检测精度、检测缺陷种类等方面均优于第1类方法。

In order to improve the intelligence of detection,this paper deals with some common tire texture defects as research object,and adopts two template-based methods to detect defects in tires.The first method uses gray level co-occurrence matrix and histogram to extract texture features from tire,the second method uses one-dimensional Gabor to extract texture feature of tire.Defects detection is realized by comparing the similarities between texture features extracted respectively from the template and test images.Experimental results show that the second method is better than first one in the detection accuracy and more adaptive for multi-type defect checking.