Applying Signal Detection Theory to investigate the quality and performance of visual inspection tasks on dynamic patterns

• Main Authors: Zi-Heng Tao, 陶志��
• Other Authors: C.Joe Lin
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/85863532691704317905

碩士 === 中原大學 === 工業工程研究所 === 95 === ABSTRACT The main purpose of the research is discussing about the back-end process of semiconductor that after testing process, in order to ensure products appearance quality, and avoid the defects flow to the customers, building up a dynamic visual inspection procedure to reduce the possibility of the defects flow to the customers while assembling and testing. Visual inspection is the same as other inspection systems; they both have rejecting good products or accepting defective products problems. The experiment of the research is divided into a senior group and a novice group. 12 operators（6 of them are senior who have 1-year work experience; 6 of them are novice who have 1-month experience.）have been surveyed; using signal detection theory, process the materials with limited and unlimited time. The affection results of mistake frequency of senior and novice operator made: While unlimited time visual inspection experiment, the senior operators have enough time to inspect materials, but there’re many “False Alarm” which cause erroneous that misjudge good products into defective products. While limited time visual inspection experiment, the speed of material movement is very fast. The senior operators inspects the material within limited time that cause many “Miss”, less “Correct Rejection”, and misjudge defective products into good products. While unlimited time visual inspection experiment, because of unlimited time, the correct ratio is higher than the correct ratio of limited time inspection. Analyzing the research, the biggest difference between senior operators and novice operators is while limited time inspection, senior operators have less “False Alarm”, on the contrary, the novice operators have many “False Alarm”. Key words: Signal Detection Theory, Vision Achievement, and Dynamic Visual Inspection.