A Composite Model Based on Shape for Fast Image Matching

Abstract

A new improved partial Hausdorff distance is presented as the accurate measurement of the degree of shape similarity between the template and images, according to this distance, a fast strategy is proposed to measure the similarity roughly. A new genetic algorithm based on fuzzy logic, which can adaptively regulate the probabilities of crossover and mutation, is used to search the optimum shape matching quickly. The composite model composed of these algorithms can finish the shape matching from coarse to fine. The experimental results show that the model is capable of the shape matching with better quality and higher speed compared with other similar matching algorithms and can be used in real-time image matching.