基于机器视觉的手机壳表面划痕缺陷检测

Surface Defect Detection of Mobile Phone Shell Based on Machine Vision

为解决手机壳表面缺陷检测采用人工目测法,检测效率低且漏检率高的问题,采用基于机器视觉的手机壳表面缺陷检测方法,实现产品缺陷的自动化检测。该检测算法采用八方向的各向异性高斯方向导数滤波器对图像进行卷积滤波,并做归一化处理;利用滤波结果图的直方图确定自适应阈值,并进行阈值分割;对图像进行细化后通过划痕缺陷长度特征进行缺陷的提取。实验结果表明,该划痕缺陷检测算法能够实现长度0.5 mm以上的划痕缺陷的准确检测,检测效率高,满足企业的实际需求。

Due to low detection efficiency and high failure rate of manual visual method for detecting the surface defects of mobile phone shell,this paper puts forward an automatic detection method based on machine vision for it.Eight directions of anisotropic Gauss derivative filter is used in the detection algorithm to do the convolution filtering of the image and normalization and the filtering result graph histogram is used to determine the adaptive threshold and threshold segmentation.The image is refined and the defects are extracted according to the scratch defect length characteristic.The experimental results show that this detection algorithm can be used to detect the scratch length more than 0.5 mm.Its detection efficiency meets the actual needs of enterprises.