Design of automated bottle defect detector for quality inspection

Ismail, Ahmad Najmuddeen (2024) Design of automated bottle defect detector for quality inspection. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full Text) Design of automated bottle defect detector for quality inspection.pdf Download (2MB)

Abstract

In the highly competitive cosmetic industry, ensuring product quality is paramount to maintaining brand reputation and customer satisfaction. One of the main issues in the cosmetic industry is to ensure the perfume plastic bottles used are free of surface and structural defects. At the industry the study is being conducted at, Tropical Bioessence Sdn. Bhd., the main issue for quality inspection of bottles is that manual inspection by humans is slow and is inaccurate. This study proposes a method based on machine vision for the identification of surface and structural defects present on multi-purpose spray bottles and sorting the defective bottles. The industry also set some requirements for the system to meet which is the system must have an accuracy of not less than 95%. The objectives are to first, design an automated bottle defect detector that can aid in identifying and sorting defective bottles as well as to analyse the effectiveness and performance of the system designed. The project is conducted within these scopes whereby a machine vision inspection system to identify surface defect on bottles is designed using MATLAB at the FTKIP lab and analysis of the system’s effectiveness and performance will be evaluated at the industry’s facilities. The inspection system utilizes image acquisition from digital cameras and the image processing and analysis is done via the MATLAB software. The USB camera will capture the digital images and these digital images will be determined by MATLAB regarding defects criteria of the bottles using black and white detection method using thresholds of 0.5, 0.4, 0.3, 0.2 and 0.1. It is determined that threshold 0.2 is the most suitable and is used for the accuracy testing. The automated bottle defect detector manages to obtain an accuracy of more than 95% for every batch inspected. However, the inspection total process is slightly slower by one minute compared to manual inspection. Therefore, to improve the speed of the system, it is recommended to use two USB cameras to speed up image acquisition process.

Actions (login required)

View Item