A new corner detection algorithm is proposed to get image corners, and then a novel improved partial Hausdorff distance is presented as the similarity measure between the template and images. Finally, a genetic algorithm with the probabilities of crossover and mutation self-adaption based on fuzzy control is used to search the optimum matching quickly. The experimental results show that the matching model composed of these algorithms is capable of fast and effective image matching and has higher speed and better quality compared with other similar matching algorithms.