Adaptive Thresholding with Iterative Fuzzy Logic-based Image Enhancements for Car Wiper Arm Defect Detection

Foo, Chi Wei (2021) Adaptive Thresholding with Iterative Fuzzy Logic-based Image Enhancements for Car Wiper Arm Defect Detection. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Foo Chi Wei.pdf Restricted to Registered users only Download (9MB)

Abstract

Defect detection is commonly applied in the manufacturing process of a product. For a product like wiper arm, the implementation of sophisticated methods for defect detection will be required due to its complex structures as well as its reflective surface. This research involves enhancements made to a previous solution by tackling the issues of misdetections of near-boundary defects and faint defects. The proposed solution is to further improve the image pre-processing stage that includes the image enhancement and thresholding steps as well as the defect detection stage that includes the blob detection algorithm. Initially, the image is to be segmented using improved Otsu-based threshold value selection to acquire the region of interest (ROI). The image enhancement step proposed involves fuzzy logic approach, which is in the form of Fuzzy Clipped Contrast-Limited Adaptive Histogram Equalisation (FC-CLAHE), where appropriate fuzzy set theory with suitable membership functions is applied, enhancing defects for better faint defects detection. As for the image thresholding step, image binarization is applied via adaptive thresholding with appropriate median filtering to eliminate salt and pepper noises from the thresholding step, while additional steps involving convex hull algorithm and probabilistic Hough transform (PHT) techniques are employed for better preservation of car wiper arm features details. The defect detection step proposed includes a multi-stage blob detection algorithm that involves blob contour analysis based on contour area and contour mean values to eliminate false-positive detections.