基于主成分分析的轮胎缺陷检测算法研究

Research on Tire Defect Detection Algorithm Based on Principal Component Analysis

针对X光射线轮胎图片中出现的灰度特征缺陷,提出基于主成分分析的轮胎缺陷检测算法.该算法首先通过图像增强技术使轮胎的缺陷特征更加明显;然后,采用主成分分析技术设置阈值,保留整幅图片的主要特征,再通过重构图像将轮胎图片映射到更少的相互正交的向量空间中,将原图与重构图像的像素值矩阵进行差分,得到初步定位的含有缺陷的图像;最后,初步定位的图像经过阈值分割和腐蚀得到只含有缺陷的图像,实现精确定位.对200幅具有不同灰度缺陷形状的X射线轮胎图片进行缺陷检测,实验结果表明:该算法能有效地检测出缺陷所在位置,且算法时效性优于其他算法.

Aiming at the gray-scale feature defects in X-ray tire images, a tire defect detection algorithm based on principal component analysis is proposed.Firstly, the image enhancement technology is used to make the defect characteristics of the tire more obvious.Then the principal component analysis technology is used to set the threshold and retain the main features of the whole picture.The tire picture is mapped to less orthogonal vector space through the reconstructed image, and the pixel value matrix of the original image and the reconstructed image is difference to obtain the preliminarily positioned image with defects.Final artwork and reconstruction of the image pixels matrix of finite difference, the pixel matrix of image contains only tire defects to have different gray level of 200 image defects in the shape of a tire X-ray images for defect detection experimental results show that the proposed algorithm can effectively detect the defect location, timeliness and algorithm is superior to other algorithms.