首页 | 本学科首页   官方微博 | 高级检索  
     

基于GPU的视频流拼接算法研究
引用本文:张燕,赵新灿,谭同德.基于GPU的视频流拼接算法研究[J].计算机工程与设计,2012,33(4):1472-1476.
作者姓名:张燕  赵新灿  谭同德
作者单位:郑州大学信息工程学院,河南郑州,450001
基金项目:国家公益性行业科研专项基金项目(200909106)
摘    要:为解决视频流的稳定实时拼接,结合图形处理器GPU强大的并行计算能力,提出了一种基于GPU的视频流拼接算法.提取视频流的帧图像,利用尺度不变特征变换(scale invariant feature transform,SIFT)算法在GPU上实现帧图像的特征提取与匹配,实现图像拼接,进而实现视频流的稳定实时拼接.基于GPU的SIFT算法充分利用了GPU的并行处理能力,加快了视频流拼接算法执行的速度,真正意义上实现了几个差异较大但具有公共视野的视频流快速稳定的拼接.

关 键 词:图形处理器  尺度不变特征变换  视频流拼接  图像拼接  特征提取

Research of video stream splicing algorithm based on GPU
ZHANG Yan , ZHAO Xin-can , TAN Tong-de.Research of video stream splicing algorithm based on GPU[J].Computer Engineering and Design,2012,33(4):1472-1476.
Authors:ZHANG Yan  ZHAO Xin-can  TAN Tong-de
Affiliation:(School of Information Engineering,Zhengzhou University,Zhengzhou 450001,China)
Abstract:To solve the stability and real-time of video stream splicing,combined with the powerful graphics processor GPU’s parallel computing capabilities,a design method of stream splicing algorithm based on GPU is presented.Extracting video stream frame image,then image stitching which contains feature extraction and matching is implemented on the GPU using SIFT(scale invariant feature transform) algorithm,to realize the stable and real-time video stream splicing.The SIFT algorithm based on the GPU makes full use of the GPU’s parallel processing capability,which accelerates the implementation of video streaming stitching algorithm and realizes the fast and stable video stream splicing with quite different but a public vision.
Keywords:GPU  SIFT  video stitching  image stitching  feature extraction
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号