Application of machine-vision to assess weight of fish (Case study: Oncorhynchus mykiss)

Abstract

Computer Vision (CV) is a relatively young discipline which has been widely used to automate quality evaluation. CV inspection of fish and fish products can provide efficient, consistent and cost effective alternative, so efforts focused on speed and accuracy of machine vision as a substitute for human inspection of foods. Machine vision is explained as the construction of explicit informative and meaningful descriptions of a physical object via image analysis. Measuring length requires catching fish from aquatic ecosystem which causes stress and mortality. Machine visions can be used as a new method for estimating weight without requiring manipulating fishes. This study tries to evaluate the relationship between weight of fish and visual features derived from image processing and present best fit relationship between weight and visual features.