A PRACTICAL FAST POINT PATTERN MATCHING ALGORITHM BASED ON MATCHING PAIR SUPPORT AND REGISTRATION CONDITION

摘要

Based on 2-D cluster approach, a fast algorithm for point pattern matching is proposed to solve the problems of optimal matches between two point patterns effectively under geometrical transformation and identify the missing or spurious points of patterns correctly. The algorithm is developed to determine the matching pair support. Experiment results prove that the algorithm is of excellent performance in correctness and speed even under certain noisy or distorted condition, and is of translation, rotation and scale invariant.