摘要
针对无人机采集图像时因为采集条件的影响,图像拼接延时较大的缺陷,提出图像匹配的快速方法,在匹配过程中引入POS观测值减少搜索时间,采用平均金字塔生成金字塔图像,特征点的匹配开始于分辨率最低的图像层,将上一层的图像匹配结果做为次层匹配的粗集。在初始的图像中得到匹配点后,基于RANSAC算法估计出变换矩阵的稳健值H;最后采用LM非线性优化算法进行优化;概述并研究图像融合的方法,实现快速拼接。实验表明,这种方法大幅提高拼接时间,提高了效率。
Unmanned aerial vehicle（uav） acquisition of the remote image of the conditions of the limit by acquisition,image mosaicing more time consuming defects.Put forward the method of fast image matching,matching process in introducing pos observation reduce search time,adopting average pyramid generating pyramid images,from the lowest resolution image layer began matching feature points,in the time a layer of matching,above the thick layer matching for value,in the original image matching point get initial used after RANSAC algorithm robust estimation transformation matrix H;Finally the LM nonlinear optimization algorithm was used to optimize the;This research and algorithm of image fusion,realizing fast joining together.
出处
《科技通报》
北大核心
2012年第10期191-192,195,共3页
Bulletin of Science and Technology
关键词
三维布料仿真
运动随机性
运动聚类
3 D simulation of cloth
sports randomness
sports clustering
