Apple stem-end/calyx identification using a speckle-array encoding pattern

Computer vision methods are widely used in apple quality grading applications. In this paper, we propose a new method to differentiate apple stem-end/calyx from true defect according to their depth information on apple surface. We design a NIR(Near Infrared) projector. The projector generates NIR structured light based on position encoding. The encoding is helpful while performing the search for the relevant neighborhood during the decoding process. The NIR structured light is used to sense the depth information of apple surface, then the stem-end/calyx region in apple image can be identified using these depth information. The identified stem-ends/calyxes are shown in the experimental results, and an overall 95.24% detection rate is achieved.