摘要
随着航空遥感视频影像应用得越来越广泛,对影像的配准速度和精度的要求也越来越高。本文利用视频影像连续重叠度大的特征,对图像进行二次配准处理,利用图像的运动估计模型,经过粗匹配和精匹配两次配准实现视频序列帧间时空配准。实验结果表明,改进的算法不仅提高配准精度,同时还提高了运算速度满足对视频图像实时处理的要求。
With the development of aerial remote sensing video image application, the speed and accuracy of image matching method need further improvement. In this paper, using the continuity and high-overlap characteristics of video image, we propose a two-time registration method. With the motion estimation model for image, we realize inter-frame spatiotemporal registration in video sequence by coarse matching and fine matching. The experimental results showed that improved algorithm has greatly improved operation speed and meets the real-time video image processing requirements.
出处
《影像技术》
CAS
2009年第2期52-56,共5页
Image Technology
基金
国家自然科学基金(60602042)资助项目
关键词
影像
角点检测
运动估计
匹配
image
corner detection
motion estimation
matching
