中文摘要：

元件图像的边缘检测是数控贴片机视觉定位系统的重要组成部分，其检测的精度与速度直接影响识别和定位的精度与速度。文中在分析元件引脚图像灰度特性的基础上，通过运用顶帽变换、边界提取等形态学的方法，提出一种精度高、实时性好的边缘检测算法。与传统和现代边缘检测算法进行实验对比，该算法检测出的元件引脚边缘在连续性和滤除孤立噪声点方面具有显著的优点。

英文摘要：

Chip image edge detection was an important part of the visual alignment system of NC place- ment machine, and the detection accuracy and the speed directly affect the accuracy and the speed of i- dentification and positioning. Based on the analysis of image grayscale characteristics of the chip lead, a high precision and real-time edge detection algorithm was given through the use of top-hat transformation and boundary extraction. Comparied with the traditional and modem edge detection algorithms, the experi- ment proved that the proposed algorithm had advanges of continuity and filtration of isolated noise point.