SIFT匹配算法在视频拼接中的优化应用  

SIFT Matching Algorithm in the Coal Mine Tunnel Video Splicing

作　　者：丁雄飞[1] 陈笋[1] 张宇[1] 张春燕[1] 

机构地区：[1]安徽大学数学科学学院,安徽合肥230601

出　　处：《宿州学院学报》2013年第4期76-78,共3页Journal of Suzhou University

基　　金：安徽大学研究生学术创新研究项目"基于SIFT算法的视频匹配拼接的研究与应用"(0100070-10117700010)

摘　　要：在视频拼接中,由于视频帧存在噪音大,光线变化、模糊以及旋转等情况,传统的SIFT算法处理后出现了较多的错误匹配点对。为了提高匹配的准确率,提出了一种优化的SIFT算法:在SIFT算法的图像匹配过程中,计算所有匹配点对所确定的直线的偏转角度,设置一个角度误差范围,对匹配点对进行角度误差约束筛选,根据得到的匹配点对对偏移量进行精确计算,实现视频帧的融合。最后通过大量实验,验证了优化后的SIFT算法更加准确地对视频帧进行匹配拼接,消除了重复区域拼接的模糊性,符合视频拼接对于偏移量计算的精确要求,实现了视频拼接的实时性和高效性。In video splicing,as video frames exist noise,lighting changes,blur and rotation, the traditional SIFT algorithm processing may have a tot of error matching points. In order to improve the accuracy of matching,an optimizing SIFT algorithm was given,and the specific method is following：in the SIFT algorithm image matching process,calculating the deflection angle of all the matching points on the line determined,setting an angle range of error, performing the angle error constraint screening on the pair of matching points ,according to the obtained matching points,calculating the offset accurately,to achieve the integration of video frames. Through a large number of experiments,the optimized SIFT algorithm more accurate in video frame matching stitching,e- liminate the repeated region stitching ambiguity ,meet the requirements ,achieve real-time video stitching and ef- ficiency.

关 键 词：SIFT算法 视频拼接 图像匹配 偏移量 

分 类 号：O244[理学—计算数学]