摘要
OLED屏的混色缺陷表现为面板上低对比度、色彩不均一的区域,具有背景整体亮度不均、各个通道灰度变化都不明显等特点,因此用基于机器视觉的方法从图像中分割出来缺陷是非常困难的。借鉴检测Mura缺陷的算法,首先用正交多项式的曲面拟合方法去构造一张背景图,再用原图与背景图相减,得到疑似缺陷的区域,然后针对混色缺陷自身的特点,提出一种基于面板像素周期性排布的检测方法,最后通过设定的阈值筛选出真正的缺陷。实验结果表明,该方法对混色缺陷的检测比较有效,且算法简单,实时性高,易于实现。
OLED chromatic defect is defined as the low contrast, uneven color of the region on the panel, with the uneven brightness of the back- ground, gray-scale of each channel does not change obviously, so segmentation defects from the image based on the machine vision method is very difficult. Imitates the algorithm of detecting Mura defect, uses the surface fitting method of orthogonal polynomial to construct a background image, and then subtracts the original image from the background image. Gets the defective area, and because of the character- istics of the chromatic defect itself, proposes a method based on the periodic arrangement of panel pixels, and selects the real defects by set- ting the threshold. The experimental results show that the proposed method is effective for detecting defects, and the algorithm is simple, real-time and easy to implement.
