Machine vision-based automatic raw fish handling and weighing system of Taiwan Tilapia

Abstract

This study proposes a vision-based automatic raw fish handling system to speed up fish cleaning and weighing. The proposed fish weighing system used a camera to capture projected images of fishes. Applying image processing techniques, physical properties of fishes, such as length, width, perimeter and area were obtained. Followed by regression analysis, weight-length, weight-height, weight-perimeter and weight-area relationships were derived. Analysis results of fifty tilapias show that coefficient of determination of the regression equation relating weight and area is 0.9303. The high value suggests that a tilapia's weight is highly correlated with its projected area. Therefore, use a tilapia's area to estimate its weight is justifiable. © 2009 Springer Berlin Heidelberg.