Shape recognition with a neural classifier based on a fast polygon approximation technique

Abstract A method is presented for the fast recognition of two-dimensional (2D) binary shapes with complicated form, like islands on a map or medical images. The proposed method is based on a new polygon approximation technique, which extracts suitable feature vectors with specified dimension, which characterizes a given shape. These feature vectors are used as inputs in an efficient neural based classifier for the fast recognition of the shape. The proposed technique is characterized by high speed performance, which is desired for real time applications.