摘要
为了解决目标跟踪中的尺度和旋转问题,提出一种基于尺度不变特征变换(SIFT)和均值漂移的目标跟踪算法。该算法首先检测模板区域和目标区域在尺度空间中的极值点,然后通过拟合三维二次函数精确定位特征点的位置和尺度,接着对目标区域和模板区域的特征点进行匹配,并根据相邻帧之间尺度和角度的连续性,去除误匹配,最后利用正确匹配的特征点中的尺度和角度信息,计算被跟踪目标的尺度和旋转角度。研究结果表明:当被跟踪目标的角度和尺度发生变化时,该算法皆具有较好的跟踪效果。
In order to solve the problem of scale and rotation in image object tracking, a novel object tracking algorithm based on Scale Invariant Feature Transform （SIFT） and Mean Shift was proposed. Firstly, scale-space extrema in the model area and object area are detected and accurate positions and scales of the feature points were located by fitting 3D quadratic fimctions. Secondly, feature points in model area and object area were matched and the mismatched points were eliminated according to continuity of scale and angle between adjacent frames. Lastly, the scale and rotation angle of the object being tracked were calculated by using the scale and angle information which was provided by the feature points correctly matched. Experimental results show that the proposed algorithm has good tracking effect no matter whether the scale or the angle of the object changes.
出处
《中南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2013年第6期2354-2360,共7页
Journal of Central South University:Science and Technology
基金
国家自然科学基金资助项目(61203375
61203083
51175261)
济南大学博士基金资助项目(XBS1241)
