Computer-aided quality control of whitefish fillets
Abstract
Computer-aided quality control of whitefish fillets is a bachelor assignment made for automation students at NTNU in Aalesund. The assignment was given by the company Optimar, a company with vast experience in creating innovative solutions for fish handling.
The task is to:
• Perform concept testing to find methods for detecting wrongly cut fillets.
• Perform concept testing to find methods for assessing blood content in a whitefish fillet.
• Include choice of camera and light source.
To solve these tasks it was done a lot of research, and several tests was performed to test out different theories. In these tests it was experimented with different types of lights and illumi- nation techniques to find a method which could highlight blood spots on a whitefish fillet. It was followed up with testing of several image processing techniques to find suitable methods for finding the area of the fillets and the blood spots.
It was created a java application that could perform the whole process form detecting the fillet to classifying it. In the application it was implemented image processing methods, and a classifying system using the K-Means clustering algorithm. The application controls an illumi- nation systemand the whole process can be closely monitored via an HMI.
The project is considered to be successful, but further work and testing is needed to make it ready for an industrial application. It was managed to highlight blood spots on whitefish fillets using one of the illumination techniques tested. The image processing methods and the clas- sification system did well on the small number of images and data accessible, and should be further worked on.