An efficient 2D deformable objects detection and location algorithm

This paper presents a complete method for the automatic detection and location of two-dimensional objects even in the presence of noise, occlusion, cluttering and/or deformations. This method is based on shape information extracted from the edges gradient and only needs a template of the object to be located. A new Generalized Hough Transform is proposed to automatically locate rigid objects in the presence of noise, occlusion and/or cluttering. A Bayesian scheme uses this rigid objects location algorithm to obtain the deformation of the object. The whole method is invariant to rotation, scale, displacement and minor deformations. Several examples with real images are presented to show the validity of the method. (C) 2003 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved.