Profiling the Wiper Arm Surface for Surface Appearance Defect Detection

Lim, Yong Chuan (2022) Profiling the Wiper Arm Surface for Surface Appearance Defect Detection. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Lim Yong Chuan.pdf Restricted to Registered users only Download (2MB)

Abstract

The wiper arm industry has a high demand for automated quality control. The defect detection in the wiper arm is complex due to the reflective surface and non-uniform structure. All related works are using an area scan camera where the image contains a narrow visible zone and low region of interest. This result in a lot of sequential images need to be taken consecutively and processed. This project proposed to use a line scan camera with an appropriate lighting system to capture the image with a wider visible zone and the defect will be detected by the blob detector. The hardware used in this project are diffused semi ring blue LEDs, line scan camera and motion platform while the software required is Pylon 6 and ASDA. The proposed method has been performed and tested on 200 images generated from 50 wipers. The results illustrate the defect detection capability of the proposed method whereby the accuracy for scratch, bump and dent is 86.16%, 97.06% and 98.02%% respectively. The capturing time and processing time per wiper is 33 seconds and 0.067 seconds respectively while the classification capability is based on 72 defective images with an accuracy of 88.89%.