摘要
针对目前无人机航空影像非同源、畸变大、处理量多的问题,提出一种改进的无人机航空影像配准方法;首先利用传统SIFT方法得到特征点,其次利用C均值聚类方法可实现准确的非监督分类的特点,对传统SIFT方法得到的特征点进行筛选,从而得出同名点;最后根据得到的同名点完成待匹配图像的投影变换完成配准;通过实验仿真证明该方法精度有较大提高,且可自适应处理不同图像,是一种有效的无人机航空影像匹配改良方法。
The UAV aerial images are nonhomologous and variety, and the processing is complex. Aiming at the problem above, an improved UAV aerial image matching method is proposed. Firstly, using the traditional SIFT method to get the feature points. Then, using the C--means clustering methods to achieve the non--supervised classification correctly. And then choose from the feature points to get the corresponding points. Finally, using the corresponding points to finish the image matching by projection transformation. Simulation and results prove that this method is higher accuracy and can adaptively process different images, making this method a effective image matching improved method.
出处
《计算机测量与控制》
2015年第6期2185-2187,共3页
Computer Measurement &Control
基金
国家自然科学基金(51307183)