Similarity Measure Using Hausdorff Distance in 2D Shape Recognition System

Abstract

Abstract-Shape plays an important role for human to recognize an object. Shape can be recognized easily by human regardless of size, rotation or translation. In this paper a shape recognition system is presented. In this research, authors investigate late improves in shape analysis and shape recognition issues with a concentrate on significant highlights of a shape. In the proposed method, the first stage is an edge detection method based on fuzzy cellular automata. In the second stage, shapes were clustered based on the differences of degree of each angle. Finally Hausdorff method is used for determining the rate of similarities. Simulation results demonstrate the system invariance to rotation, translation and scale.