A modified Hausdorff distance using edge gradient for robust object matching

Conventional object matching algorithms based on Hausdorff distance mostly use edge position information to compute distances. In this paper we represent an edge point using its position and the strength of its gradient to define a 3D distance function. We propose a modified Hausdorff distance based on this 3D distance function, and investigate its application in object matching problems. Finally, experimental comparisons are performed with conventional methods. At this stage, we tested different methods on the images with different levels of noise. Their sensitivities to geometric distortion are also evaluated against image rotation and scale change. Experimental results show that the proposed algorithm is more robust and reliable.