A shape classifier based on Hopfield-Amari network

The representation and recognition of a planar shape based on contour information is an important issue in computer vision. We propose a method for extracting the main features of a contour using the curve bend function (CBF), which can be used to characterize the contour completely. A Hopfield-Amari network is built based on the CBF description to perform classification of planar shapes. The experimental results demonstrate that the proposed system is powerful and reliable for solving shape recognition problems.