【摘要】 传统的图像匹配方法往往会因为匹配图像与模板图像之间存在任意角度旋转而失效,而现有的抗旋转图像匹配方法因为计算量大而不能满足实时应用的需要。因此研究一种旋转无关的图像匹配算法成为必要。将圆投影匹配算法结合多值模板和塔形加速技术加快了滑动窗口的搜索效率,并使用伪Zernike矩进行精匹配。实验结果表明,改进后的算法在保证匹配结果准确性的同时,大大提高了运行的实时性。更多还原

【Abstract】 Traditional image matching method would fail because of the rotation between reference and template images.But existing rotation invariant matching method can not meet the demand of real time using because of the high computation cost.In this case,the research of rotation invariant scene matching method is necessary.In this paper,the author combined circular projection matching method with multiple templates and pyramid images technologies,inorder to speed up the sliding of the search window.Pseudo Zernike moment is used to get accurate matching position.The experimental result shows that the algorithm proposed in this paper speeded up the execution a lot while keep its validity and robustness.更多还原