Study of Character Recognition and Defect Inspection of IC Laser Marking Using Back-Propagation Neural Networks Method

• Main Authors: Shih-Hsien Wang, 王世憲
• Other Authors: SHIU-HSUAN CHIU
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/75853008571618266439

碩士 === 國立臺灣科技大學 === 自動化及控制研究所 === 98 === This thesis presents a recognition system that contains optical character recognition (OCR) and automatic optical inspection (AOI) proceedings for laser marking of IC system. The recognition system can be divided into three flows: IC location alignment, characters extraction and recognition, and defects inspection. For the OCR system, eighty image features are designed and applied to the Back-Propagation Neural Network (BPNN). The experimental rusts show that the OCR system can achieve 100 of correct recognition rate. For the AOI system, we use four image process methods to detect six kinds of defects for the IC laser marking. From the experimental result, the correct detection rate is 96.3% achieved. In addition, the time required for the OCR and AOI proceeding is averaged about 0.189 seconds.