Duck Egg Sexing by Eccentricity Determination Using Image Processing

Abstract

Manual duck egg sexing is practiced in the Philippines for a long time now. The method used was established on the proposition that the male duck eggs are elongated while the female eggs are more rounded. This paper proposes for the use of eccentricity to measure the roundedness of the egg. It aims to establish the accuracy of egg eccentricity in determining the sex of fertilised duck eggs. A total of 103 egg samples are considered in this study. A two-dimensional still picture of each egg sample is taken and enhanced before the eccentricity is determined. The eccentricity of each egg is computed using Matlab, a numerical computing software. After determining the eccentricity of each egg sample, a specific eccentricity threshold value is determined to separate the male from the female. Any egg sample which eccentricity is falling below this threshold value is considered a female egg, while any egg with eccentricity value equal to or higher than the threshold value is a male egg. The accuracy of the proposed method is based on the actual sexing done on the hatched eggs using the vent sexing method. Results show that with an eccentricity threshold value of 0.6441, up to 86% accuracy is attained in predicting the sex of duck eggs using the proposed method. This accuracy has a high significance in the segregation of female and male eggs in the duck egg industry in the Philippines and other parts of Southeast Asia.