Shrimp Shape Analysis by a Chord Length Function Based Methodology

Abstract

The most expensive operational cost in shrimp farming and in every aquaculture system is feeding. To estimate the quantity of food is necessary to know the total biomass of the pond. Traditionally, this is done by taking samples and weighting, which is invasive and stress the animals. Non intrusive methods have been tried to estimate pond biomass using different technologies, being one of them computer vision. Computer vision faces several challenges, such as the problem of how to identify shrimps, count them, estimate their size and their mass. In this work, a chord length function based methodology is proposed as a viable alternative to analyze shrimp’s shape and count them, this methodology generates histograms of the shape of the shrimps and therefore, a set of statistical parameters (mean, median, mode, variance, standard deviation, maximun and minimum) to quantify shape and which can be useful to identify shrimps, estimate their sizes, and even find a relationship between morphometric measures with respect to biomass.