The Recognition and Capture of Wafer Probe Marks in IC Package Using Computer Vision
| Autor: | Liu, Ping-Ju, 劉秉儒 |
|---|---|
| Rok vydání: | 2017 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 105 Semiconductor is a highly automatically industry, but semiconductor companies rely on engineers using eyeball analysis to judge wafer defect, like probe mark inspection during chip probe (CP). The probe mark will reduce wire bonding strange at chip assembly and packing stage, then cause final product test failed. Some of researcher had found the way of probe mark recognition and classifying, however, it's not close practice online applications enough. To fill the gap, the study aims to develop an inspection software by using computer programming language Python and extend library Open CV, SciKit-learn for machine learning, and compare the model performance of artificial neural network (ANN). The results show SVM has better test precision, recall and F1-score than artificial neural networks, and compute time is much shorter. This will help to reduce online error judge cost. Lower hardwarerequirement and good model performance are also the main benefits for inspection hardware design. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|