Multi-dimensional Wiper Arm Defect Detection Using Genetic Algorithm

Ler, Wei Xian (2021) Multi-dimensional Wiper Arm Defect Detection Using Genetic Algorithm. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Ler Wei Xian.pdf Restricted to Registered users only Download (1MB)

Abstract

there has been an increasing demand for the application of automatic inspection systems for quality products to replace manual labour in most manufacturing industries including car wiper arm manufacturing industry. Traditional inspection system relies on human vision checking which leads to high labour cost and other inaccurate detection issues. Therefore, this paper presents a multi-dimensional wiper arm defect detection system using Genetic algorithm. This defect detection system employs an image processing technique to inspect the scratch, bump and dent defects on the wiper arm surface. The proposed method includes pre-processing, feature extraction and defect detection. Pre-processing widens the visible zone of wiper images by using the image stitching method. Various kind of image stitching approaches are studied and analysed in this paper. Then, Genetic algorithm is used to enhance the feature of the defects because the features of the wiper are blurred after image stitching. Genetic algorithm is an optimization approach which is useful in image enhancement. Image thresholding is applied to identify the defects in binary images. The defect detector using the contour method locates the defect on the image. The proposed method has been performed and tested on 136 images generated by 264 original wiper arm images. The results illustrate the capability of the proposed defect detection method with an accuracy of 90.51 % and a processing rate of 0.218s per image.