一种适用于玻璃生产线的瑕疵检测算法

An algorithm of glass defection on-line detection system

以一个基于线阵CCD的玻璃生产线检测系统为硬件平台,重点研究了玻璃瑕疵的检测算法。首先提出了卡尔曼滤波器跟踪背景光强变化的方法,利用瑕疵在时间轴上造成的灰度突变来识别瑕疵的到来与结束,然后通过形态学的方法对瑕疵点的坐标进行定位,最后将玻璃瑕疵的位置坐标及其图像送至上位计算机进行分类处理。本文的算法已经固化于FPGA构成的实时处理系统中,实现了精度高达0.1mm的玻璃瑕疵实时检测。

This paper based on a hardware platform of glass defection on-line detection system,it mainly discussed the algorithm of glass defection on-line detection system.Kalman filter is proposed to trace the change of background light intensity.Glass defection is recognized according to the grey variation which changes with the time.A morphology method is used to locating the coordinate of glass defection.The image and coordinate of the glass defection is finally delivered to the computer,it is then sorted by the computer.The algorithm of this paper is applied in a real time processing system that is based on FPGA.The accuracy of this glass defection on-line detection system reaches 0.1mm.