Machine Vision Based Weight Prediction for Flatfish

In fisheries industry, flatfish accounts for about half of the domestic fish production and is a specie with active aquaculture activities. As flatfish grows, classification according to the size should be periodically carried out for efficient growth. In this paper, the correlation between area and weight of flatfish required for classification algorithms using machine vision is derived. 120 simulations are performed using flatfish models and the area is obtained through image processing. In image processing process, the area of flatfish model is calculated using a reference square known in size. The weight of actual fish corresponding to the length of the model is obtained using the reference equation. Then, regression analysis is performed to derive the interpolation of linear and power equations which show correlation between area and weight of the flatfish.