基于BSIFT的无人机影像快速拼接算法

UAV Image Quick Mosaic Algorithm Based on BSIFT

针对传统SIFT算法在无人机影像配准过程中速度较慢的问题,提出了一种改进的BSIFT影像拼接方法。针对SIFT特征点描述符计算复杂、占用存储空间过大、匹配耗时过长的缺点,使用BRIEF特征点描述符代替,利用改进方法(BSIFT)对影像进行特征点的提取与匹配,利用RANSAC算法剔除错配点并计算变换矩阵。实验结果表明:本文提出的方法提高了影像拼接速度,而且提高了拼接精度。

This paper proposed a method BSIFT to deal with the low image registration speed problem of the traditional SIFF algo- rithm. Because SIFF feature point descriptor has computation complex, too much storage space, matching time - consuming long faults, we use BRIEF to replace it. The BSIFF algorithm is introduced to extract the feature points of UAV images and match them, RANSAC is applied to remove the mismatched points and estimate the parameters of the transformation matrix. The experimental resuits show that the image registration time is reduced obviously and the stitching accuracy is also improved.