为了避免图像配准精度对液晶显示屏(LCD)线路缺陷检测准确率的影响, 采用一种基于图像轮廓分析的新方法, 基于深度优先搜索寻找图像轮廓, 并根据格林公式计算轮廓面积, 将待检测LCD的线路轮廓面积与标准模板的轮廓面积比较, 从而判断是否存在短路、断路、孔洞和孤岛缺陷。该方法不需要图像配准, 降低了算法精度要求, 从而提高了检测的正确率。通过对200片小型LCD的测试, 检测准确率达99%。结果表明, 该方法可以快速正确地检测出图像上的所有轮廓, 并计算出轮廓面积, 用面积比较代替传统的图像像素比较, 检测短路、断路等缺陷的正确率有了较大提升。该算法在LCD线路缺陷检测方面具有很好的应用前景。

In order to avoid the impact of image registration precision on detection accuracy, a novel approach for circuit defect inspection on liquid crystal displary(LCD)panel was adopted based on image contour analysis. The contours were found by depth first search(DFS), and their areas were calculated according to Green formula. In order to ascertain whether there are any defects such as short, open, hole and island on this LCD, the circuit contours' areas of the LCD to be detected are compared with the areas of standard template. This method needs no image registration, therefore, precision requirement for algorithm is reduced and detection accuracy is improved. Through the experiment on 200 small LCDs, the accuracy comes up to 99%. The results show that all contours are found and their areas are calculated correctly and rapidly. With comparison of areas instead of images, the accuracy of detecting defects such as short and open circuit is improved. This method has great application future in the field of defect detection for LCD circuits.