Detecting quality defects in production using image processing method

In this thesis, the feasibility of automating the quality control of mechanical nut on sheet metal parts used in automotive production lines using image processing methods has been investigated. Specifically, the aim is to inspect the accuracy of the mechanical nut sizes and positions, detect foreign objects, and identify missing or defective mechanical nut. Using the OpenCV library for image processing, mechanical nut and foreign objects were detected through an Hue,Saturation,Value (HSV) color space-based method, and the number and sizes of nuts in defined regions on the parts were analyzed. The proposed system in the thesis processes real-time images captured by cameras to check whether the mechanical nut comply with the desired dimensions and layout. Additionally, it detects foreign objects that may affect quality outside the defined areas. This system aims to minimize human errors in quality control processes and enhance production efficiency. By automating the mechanical nut inspection on parts using image processing techniques, the system reduces errors encountered in manual inspections and helps the production line run more smoothly Keywords: Image processing, Quality defects, OpenCV