汽车轮毂缺陷检测中的缺陷增强与字符去除技术

传统的轮毂图像缺陷检测算法,直接对采集到的图像进行检测,准确度低是其主要缺点。鉴于此,提出了一种汽车轮毂缺陷检测中的缺陷增强与字符去除技术。该技术通过对原图像进行数学形态学处理得到伪模板图像,然后将其与原图像相减的结果进行灰度拉伸,再叠加到原图像上以增强缺陷特征,最后使用双边滤波技术去除轮毂字符信息。实验结果表明,使用缺陷增强和字符去除技术之后,在满足时间限制的前提下,轮毂缺陷检测算法的准确度得到了提高。

Defect Enhancement and Character Removal Technology of Defect Detection of Automobile Hub

The traditional defect detection algorithm of hub images detect the collected images directly, low accuracy is its main disadvantages. In this paper, defect enhancement and character removal technology of defect detection of automobile hub is proposed. First of all, the original image is processed to get pseudo template image by mathematical morphology. Secondly, the pseudo template image minus the original image, and the result image is processing by gray stretching. Thirdly, the result image is added to the original image to enhance the characteristics of the defect. At last, bilateral filtering technology is used to remove the character information in the hub images. Experimental results shows, under the premise of limited time, accuracy of detection algorithm of hub defect is improved after using defect enhancement and character removal technology.