融合颜色直方图及SIFT特征的自适应分块目标跟踪方法

Adaptive Fragments-based Target Tracking Method Fusing Color Histogram and SIFT Features

针对目标跟踪过程中目标颜色相近、尺度变化及遮挡等问题,该文提出一种自适应分块并融合颜色直方图及SIFT特征的目标跟踪方法。自适应分块采用目标颜色投影和成像角度作为分块标准,使各子块具有一定相异性并保证分块数目;各子块使用颜色直方图和SIFT特征描述,通过计算SIFT特征点的尺度变化自适应地改变跟踪窗口尺度;在跟踪过程中对子块的权重及相应模板及时更新,当目标表观变化较大时重新对模板自适应分块。实验表明,该方法能准确有效地跟踪目标,并在颜色相近目标跟踪、尺度自适应及遮挡处理等方面具有较好效果。

In order to solve the problems of color similar target, scale change and occlusion in target tracking, this paper proposes an adaptive fragments-based target tracking method fusing color histogram and SIFT features. Taking color projection and imaging angle as the standard, the adaptive fragment method can ensure not only the distinctive features of fragments but also the number. Each fragment is represented by color histogram and SIFT features. Adaptive scale selection can be obtained by computing the scale change of SIFT features. The weighting factor and corresponding template of each fragment are updated during tracking and the whole target template will be divided again if the target appearance changes significantly. Experimental results demonstrate that the method can track the target accurately and effectively and has advantage in color similar target tracking, adaptive scale selection and occlusion handling.