Study of Target Face Search Algorithm for Video Advertisement
暂无分享,去创建一个
Retrieving the wanted contents within large amount of video quickly is one of the key technologies of advertisement recognition. This paper analyzed the principle of AdaBoost algorithm and SURF matching algorithm, and designed a system of target face retrieving. The Adaboost algorithm was used to detect the faces from video key flames, and SURF matching algorithm was used to match the targeted face in a video. This system especially solved the matching problem for the changed targeted face brought by the different angle of view and scale in the video. This work was programmed by C# language in the environment of Visual Studio 2010 and EmguCV. The experimental results showed that the system could retrieve the appearing time of the target face accurately, and it was easy to be operated and was with a certain robustness and practicability.
[1] Andrew Zisserman,et al. Identifying individuals in video by combining 'generative' and discriminative head models , 2005, Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1.
[2] G LoweDavid,et al. Distinctive Image Features from Scale-Invariant Keypoints , 2004 .
[3] Andrew Zisserman,et al. Person Spotting: Video Shot Retrieval for Face Sets , 2005, CIVR.
[4] Matthijs C. Dorst. Distinctive Image Features from Scale-Invariant Keypoints , 2011 .