碩士 / 中原大學 / 通訊工程碩士學位學程 / 103 / The objective of this study is to optimize the detection process of defects during quality control of lures by developing a system using machine vision to enable automatic inspection of lures defects. The goal is to facilitate the elimination of human error and providing an efficient and reliable method to conduct quality control. This paper focuses on two kinds of defects, the presence of bubbles in the product and oil staining. Due to the translucent nature of the lure, back lighting is used to capture the image. Image processing is divided into two processes; image correction and blog segmentation. For image correction, binary image is use to find the direction and to compute the rotation angle for the lure, linear regression is used to fit a predictive model to an observed data, enabling a prediction of the Lures direction. For blog segmentation, canny edge detection and contour assessment are used to detect the position of the frame. To detect a bubble defect, image contrast enhancement is used to find the local minimum value of the segmented defect area. To detect an oil stain defect, a grey-scale is used to determine if the affected area exceeds the by the predetermined maximum value set by the company. System simulation were conducted by testing 1000 images for both defect types, results showed that the rate detection of bubble defects was 92.8 % and the rate detection of oil stains was 97.8 %. It is concluded that the automatic inspection system provides a reliable method to conduct quality control and that the system could be used by the company.
| Identifer | oai:union.ndltd.org:TW/103CYCU5650070 |
| Date | January 2015 |
| Creators | Kai-Cheng Chuang, 莊凱丞 |
| Contributors | Li-Der Jeng, 鄭立德 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 84 |
Page generated in 0.0081 seconds