earticle

영어

This paper presents algorithms based on image processing for detecting internal blood spots and eggshell dirt by processing acquired images from eggs under different illuminations. The algorithm can also detect the severity of dirt on eggshell. In order to carry out image processing and extract useful features of captured images of eggs by machine vision we developed an algorithm in HSI color space. The hue histogram was used for blood spots detection, and maximum values of two ends of histogram were selected as criterions of defect detection. Eggshell dirt was detected using connected areas detection technique. The results of experiments showed that accuracy of differentiation of blood spots algorithm was 90.66% of defected eggs and 91.33% of intact eggs and total average of this algorithm was 91%. Accuracy of differentiation of dirt detect algorithm was 86% of clean eggs, 83% of low dirt eggs and 88% of high dirt eggs. Then total average of this algorithm was 85.66%.