A Textural Distributions-Based Detection Of Hazelnut Axial Direction

Since the shell of the hazelnuts is very hard, it is necessary to squeeze the slit in the axial direction to facilitate peeling. When the hazelnuts are automatically processed in the industry chain, the axial of the hazelnuts needs to be quickly positioned. In this paper, the sum of projection gradient in the sensitive area of the hazelnut image is used to locate the hazelnuts axial. Firstly, a search template with discrete paths is established to find the hazelnut contour and extract the hazelnut region from the image. Secondly, the sensitive area is selected to get projection histogram at diffident orientations, and then the gradient sums of the projection histograms are calculated. Thirdly, the axial orientation of the hazelnut is determined with the biggest sum. The experiments results show that the projection gradient sum method is fast enough and can meet the requirements of industrial production. The location accuracy of the projection gradient sum method is 94.2%. (C) 2021 The Authors. Published by Atlantis Press B.V.