A thinning-based algorithm to characterize fruit stems from profile images

In this paper we present a thinning-based approach to detect protrusions from binary patterns, and its use to characterize fruit stems from fruit profile images. The method to detect protrusions in binary patterns is based on the property of iterative thinning algorithms which causes protrusions to become skeletons during the first iterations of the thinning process. A modified thinning method, and a restoration of the thinned parts using criteria based on some topological properties of digital images, allow to identify, locate and characterize in size and length these protrusions from binary patterns. The method has been applied to characterize fruit stems, allowing the successful detection of a wide range of stem shapes, with high effectiveness, 99% of stems being correctly identified.