Abstract: This paper proposes an edge detect method which combines with LoG operator and mathematic morphology for glass fragment.This method is established on a rapid transmission system which carries many kinds of glass fragment.The system is required to have rapid edge detect for the glass fragment on the conveyor belt，so the image processing speed must be quick and accurate.According to studies，the LoG operator has less calculated amount and high efficiency of edge detection，but is sensitive to noises.The mathematic morphology can make up this drawback exactly and is easy for hardware implementation.The experimental result indicates that the edge detect effect is good while using dilate and erode operation of mathematic morphology firstly and then the LoG operator for smoothing the images spurred by noises.

Key words: LoG operator, mathematic morphology, glass fragment, edge detection

摘要： 提出了一种LoG算子与数学形态学结合的玻璃碎片边缘检测方法，该方法是建立在一个承载着各种各样玻璃碎片快速传输系统上的。这个传输系统需要对传送带上的玻璃碎片进行快速的边缘检测，因此要求图像处理速度既快又准。经过研究，LoG算子计算量少，边缘检测效率高，但对噪声敏感;而数学形态学恰恰能弥补这一缺点，而且易于硬件实现。实验结果表明先用形态学中的膨胀和腐蚀操作对含噪声的图像进行平滑处理，再用LoG算子进行边缘检测的效果很好。

关键词: LoG算子, 数学形态学, 玻璃碎片, 边缘检测