基于机器视觉的导光板漆面缺陷检测方法及实现

Machine Vision-Based Detection Method and Implementation of Defects on Light Plate Paint Surface

针对导光板在生产过程中人工检测效率低的问题,提出了一种基于机器视觉的导光板漆面缺陷检测方法。根据划痕、起泡、暗点、色差等各类典型缺陷的特点设计了对应的检测流程。改进了对比度调整算法,提出了一种以Sigmoid函数为映射曲线的自适应灰度变换方法增强起泡、划痕等点状、线状缺陷特征,采用局部自适应直方图均衡化算法突出色差缺陷轮廓,最后结合形态学处理、边缘检测等算法实现了对应缺陷的检测。实验结果表明该方法对各类点状、线性、色差缺陷的检出率达到90%。

This paper proposes a machine vision-based method for detecting defects on the paint surface of a light plate,which addresses the low efficiency of manual detection in the production process.Corresponding detection processes were designed based on the characteristics of various typical defects,such as scratches,bubbles,dark spots,and color differences.This paper improves the contrast adjustment algorithm and proposes an adaptive grayscale transformation method with a Sigmoid function as the mapping curve to enhance point and line defects such as bubbles and scratches,and a local adaptive histogram equalization algorithm is used to highlight the color difference defect areas.Finally,with the combination of morphology processing,edge detection,and other algorithms,the detection of corresponding defects was achieved.The experimental results show that this method has a detection rate of 90%for various types of point,line,and color difference defects on the light plate.