用机器视觉检测集成电路芯片的图像分割方法研究

Image segmentation method research in using machine vision to inspect IC chips

在研制检测集成电路芯片的机器视觉系统中,研究运用Mahalanobis距离方法,对摄取的芯片图像进行图象分割。实验结果表明,对处于环境复杂、低对比度、细节多的芯片图象能够产生很好的分割效果。该方法鲁棒性强,能保证分割质量,能显著提高机器视觉的检测精度。亦可用于对其它产品的尺寸及表面质量检验测量。

In a IC chip inspecting machine vision system, we used Mahalanobis distance method to segment the captured chip image. Experi ment Results show that the method performs well towards the images featuring complex-environment, low-contrast and multi-detail. With strong robustness and appealing segmentation quality ,the method can be used to consid- erably improve the inspection precision. It can be used to perform dimension and surface quality inspection of other products as well.