On-line detection of defects on fruit by machine vision systems based on three-color-cameras systems

Abstract

How to identify apple stem-ends and calyxes from defects is still a challenging project due to the complexity of the process. It is know that the stem-ends and calyxes could not appear at the same image. Therefore, a contaminated apple distinguishing method is developed in this article. That is, if there are two or more doubtful blobs on an apple’s image, the apple is contaminated one. There is no complex imaging process and pattern recognition in this method, because it is only need to find how many blobs (including the stem-ends and calyxes) in an apple’s image. Machine vision systems which based 3 color cameras are presented in this article regarding the online detection of external defects. On this system, the fruits placed on rollers are rotating while moving, and each camera which placed on the line grabs 3 images from an apple. After the apple segmented from the black background by multi-thresholds method, defect’s segmentation and counting is performed on the apple’s images. Good separation between normal and contaminated apples was obtained for threecamera system (94.5%), comparing to one-camera system (63.3%), twocamera system (83.7%). The disadvantage of this method is that it could not distinguish different defects types. Defects of apples, such as bruising, scab, fungal growth, and disease, are treated as the same.