改进SIFT小型无人机视频序列图像自动拼接方法  

An automatic mosaic method of unmanned aerial vehicle video serial images based on improved SIFT

作　　者：熊自明[1,2] 闫鹤[1] 

机构地区：[1]信息工程大学测绘学院,郑州450052 [2]国际关系学院,南京210039

出　　处：《测绘科学》2013年第5期133-136,共4页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(40971239)

摘　　要：针对小型无人机视频序列图像视点离散、视角变化有一定运动规律的特点,本文结合Harris特征点和SIFT特征向量的优势,研究实现图像的自动配准;并采用光度对准的方法和加权平均的方法消除光度差异,较好地实现了视频序列图像的无缝拼接。实验证明该方法在图像拼接的准确性、效率等方面较经典的SIFT算法有较大提高。Due to the disperse view points and regular movement of view angle of UAV video serial images, this paper studied to realize the automatic mosaic for the images. Then the photometric registration and weighting-average blending were used to smooth and improve the result image for the seamless splicing. The results of experiment proved that the way of mosaic was more efficient and exact than the classic SIFT algorithm in image mosaic.

关 键 词：无人机视频序列图像 图像配准 HARRIS SIFT PCA BBF PROSAC 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]