Conventional and deep feature oriented quality inspection of internal defected eggs using infrared imaging

Abstract

Automatic separation of defective eggs from qualified ones would lead to a great reduction on the graders visual stress as well as to an improvement on the quality control process. Due to the increasing incidence, Infrared Imaging Technology provides an important window for eggs sorting especially when the defects are not visible externally. In this paper, we have investigated the role of infrared imaging for classification of internal defective eggs from the fresh eggs. For our work, a new infrared image dataset of fresh and defective eggs (i.e. internal defective) has been designed by maintaining standard acquisition protocol. The study also includes investigation of conventional and deep features for accurate separation of defective eggs from qualified ones. Experimental results shows that DeepFS outer performs the remaining three feature sets (i.e. AComFS, SSigFS and ASigFS) with an average accuracy of 96.26% for all the used classifiers and hence able to effectively separate the fresh and defective eggs.