计算机与数字工程2017，Vol.45Issue(3)：527-530,557,5.DOI:10.3969/j.issn.1672-9722.2017.03.025
一种改进的视频序列SIFT特征检测方法
An Improved Method of Detecting SIFT Features in Video Frame
陈滨 1赵建军 1马同刚 2王远强2
作者信息
- 1. 海军航空工程学院兵器科学与技术系 烟台 264001
- 2. 91899部队 葫芦岛 125001
- 折叠
摘要
Abstract
Although SIFT algorithm has great advantage of extracting the invariant features, it is a challenge to use the algorithm real-time because of the time complexity.The paper improves the SIFT algorithm by limiting the area of extremes detection.The radius of the area is updating with the global motion vectors between the continuous frames of a video.With the improved algorithm, the images of the areas are obtained in which the extremes to be detected.The algorithm keeps the amount of SIFT features and is proved to be more efficient by decreasing the time cost by extremes detection.关键词
SIFT算法/全局运动矢量/视频图像/尺度空间Key words
SIFT algorithm/global motion vector/video frame/scale space分类
信息技术与安全科学引用本文复制引用
陈滨,赵建军,马同刚,王远强..一种改进的视频序列SIFT特征检测方法[J].计算机与数字工程,2017,45(3):527-530,557,5.
