Abstract

Tao Zhu and Qiaohua Wang

A method of non-destructive detection of Sudan dye duck eggs was developed using image processing and fuzzy cluster analysis. Duck egg color images were obtained by using a computer vision device. Through the component analysis of RGB image, it was found that the yolk region could be viewed distinctly in the gray scale image of B-component, and this characteristic was used to separate the yolk region from the white region. By extracting and comparing the color parameters, the R-component values of the yolk region showed obvious differences for Sudan dye and natural red-yolk duck eggs. A fuzzy discriminant model for the detection of Sudan dye duck eggs was established using the yolk’s color parameters. The experimental results indicated that this model had good capability of identification for Sudan dye eggs.