CAJ | 학술논문

Mean Shift算法在视频序列中的目标跟踪已经广泛被应用于计算机视觉研究以及应用中。该算法应用于跟踪中,具有计算量低,可实时跟踪等优点,但有时会出现漂移问题。本文针对传统Mean Shift算法的实现进行了研究和总结。并提出了基于分块的改进方法,在Mean Shift跟踪算法中加入了空间信息。实验证明改进算法与原始算法比较具有更好的跟踪精度。
Mean Shift산법재시빈서렬중적목표근종이경엄범피응용우계산궤시각연구이급응용중。해산법응용우근종중,구유계산량저,가실시근종등우점,단유시회출현표이문제。본문침대전통Mean Shift산법적실현진행료연구화총결。병제출료기우분괴적개진방법,재Mean Shift근종산법중가입료공간신식。실험증명개진산법여원시산법비교구유경호적근종정도。
Object tracking in video stream using Mean Shift Algorithm has been widely employed in Computer Vision Research and Application.This algorithm has the feature of low computation complexity and the ability of real-time tracking,while it also has the problem of shift.In this paper,classical Mean Shift Algorithm has been researched and summarized,and a new method that partition the big block into small pieces is asserted,which add spatial information into the Tracking Algorithm.Some experimental result prove that the improved algorithm is more accurate than the classical one.