面向视觉伺服的芯片图像Zynq预处理系统

Chip Image Preprocessing System with Zynq for Visual Servo

摘要:

晶圆芯片定位精度直接影响到芯片制造的效率和质量，视觉伺服作为一种实时控制方法，引入了视觉反馈，不仅能够保证芯片定位精度还具有良好的鲁棒性。同时图像处理的实时性对视觉伺服至关重要。传统的使用PC机和网线的图像处理方式实时性和准确性较差，为了提高这一过程中视觉反馈的精度和实时性，提出并设计了一套面向视觉伺服的芯片图像Zynq预处理系统。系统通过在Zynq PL端的片内设计和PS端的控制设计，对采集到的芯片图像进行去噪、Sobel处理，并给芯片粗定位，为芯片的伺服控制提供有效数据。经实验表明，该系统实时性强，位置识别稳定，达到设计要求。

Abstract:

The positioning accuracy of wafer chip directly affects the efficiency and quality of chip manufacturing. As a real-time control method, visual servo introduces visual feedback, which not only ensures the positioning accuracy of chip but also has good robustness.At the same time, real-time image processing is very important for visual servo. The traditional image processing method using PC and network cable is poor in real-time and accuracy. In order to improve the accuracy and real-time performance of visual feedback in this process, a Zynq preprocessing system for visual servo-oriented chip images was proposed and designed. Through the on-chip design of Zynq PL terminal and the control design of PS terminal,the system performs de-noising,Sobel processing on the acquired chip image,and rough positioning of the chip to provide effective data for the servo control of the chip. The experiment indicates that this system has well real-time performance and provides steady position recognition,which meets the design requirements.