DESIGN AND IMPLEMENTATION OF IMAGE PROCESSING ALGORITHM AND HMI DISPLAY ON AUTOMATIC VISUAL INSPECTION SYSTEM

• Main Author: Fauzi, David
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/67878
• Institution: Institut Teknologi Bandung

PT Paragon Technology and Innovation is one of the largest FMCG companies in Indonesia that focuses on manufacturing cosmetic products. The manufacturing process has a very tight packaging schedule that has been determined by the Planner. Due to the high production target and the large number of SKUs produced on the same packaging line, there is possibility of errors such as mismatch of variant information on secondary packaging, which could lead to an increase in Cost of Poor Quality and a decrease in customer trust in the brand. An automated visual inspection system is implemented to detect mismatchesin variant information in the product’s secondary packaging and reject defect products. To capture packaging image, 3 cameras are mounted on a cartoning machine with a field of view 36° - 66° and a minimum illumination intensity of 138 Lux which is capable of performing image acquisition with a resolution of 1920 x 1080 pixels and a bit depth of 24 bits in real time. Based on the test results, the image processing algorithm can classify packaging quality with varying degrees of accuracy in the range of 77% to 96% and a maximum inspection processing duration of 179.83 ms. The level of accuracy and processing time are influenced by various factors such as packaging design and features detected in the image processing step. The HMI panel is used to configure the inspection algorithm parameters that have a SEQ rating of 6 and have fulfilled all the required functionality. Further development of image processing algorithms is needed to achieve better inspection accuracy and faster processing time. It is hoped that the automated visual inspection system can assist the Quality Control process on products in a manufacturing environment.