摘要
图像匹配是计算机视觉的关键技术之一,图像匹配的速度和匹配的可靠性、精度一样,也是性能的重要体现。由于传统图像匹配算法是遍历性的,匹配速度较慢,因此在实际应用中受到限制。本文把自适应遗传算法与归一化积相关相结合实现图像的快速粗匹配,然后在粗匹配点的邻域内利用相位相关算法实现图像的精匹配。本文算法不但能保证匹配的精度,也能大大提高匹配的速度。在实验中,与传统的序贯相似性检测算法(SSDA)作了比较,证明了本文算法的有效性。
Image matching is one of one key technology of computer vision. The speed of matching is as important as the dependability and precision of matching. Because of the ergodicity of conventional image matching algorithm, the speed of matching is slower. It is limited in practice. A hybrid algorithm combining adaptive genetic algorithm and unitized product correlation realizes the fast and coarse image matching. Then we can utilize phase correlation to realize the precise image matching in the neighborhood of the point of the coarse image matching. We can obtain the precise matching and improve the speed of matching by using the algorithm. In our experiment, it is proved to be effective after comparing with the conventional sequential similarity detection algorithm (SSDA).
出处
《浙江工业大学学报》
CAS
2003年第6期599-603,共5页
Journal of Zhejiang University of Technology
