Starfruit Shape Defect Estimation Based on Concave and Convex Area of a Closed Planar Curve

In this paper, a shape representation based on concave and convex area along a closed curve is presented. The proposed technique involves the process of the curvature estimation from the input curve and search for its corresponding critical points. By splitting the critical points into concave and convex categories, the concave and convex area are computed. From these statistical features, two problems related to the shape (curve) are investigated. Here, the proposed technique is tested on shape defect estimation and shape recognition of starfruit. In the first case, defect is measured by computing concave energy, which is proportional to the defect. For shape recognition, starfruit’s stem is identified and removed from the starfruit’s shape, as it will contributes false computation of defect measurement. For both cases, the proposed technique is tested with three different curvature estimation techniques to validate the results