EDGE EXTRACTION ALGORITHM FOR FEATHER SEXING POULTRY CHICKS

ABSTRACT An investigation of methods to automate poultry chick sexing was performed and, as a result of the study, an algorithm for machine vision feather sexing was developed. The algorithm uses edge detection techniques applied to digitized images of the wing tips of feather-sexed chicks. Feather shaft lengths extracted by this method were then used to develop sample statistics to classify an individual image. A data set of 200 images (100 male and 100 female) was constructed, and various statistical tests were investigated to determine the accuracy of the algorithm for the sexing operation. Accuracies ranged from 50% to 89%, with the highest obtained from discriminant analysis. Several improvements are possible which may make such a technique a viable process for automating the sexing operation.