基于颜色属性直方图的尺度目标跟踪算法研究

利用目标颜色信息的跟踪算法，容易受到环境光照、尺度变化、相似背景等因素的干扰，导致跟踪任务失败。为了克服以上问题，该文提出一种基于颜色属性空间的鲁棒尺度目标跟踪算法。该算法首先将原始的RGB颜色空间映射到颜色属性(Color Names, CN)空间，减少目标颜色在跟踪过程中受环境变化影响。然后采用一种背景加权约束的颜色属性直方图，来抑制相似背景的干扰。最后，为了解决目标尺度变化带来的影响，先用梯度上升法粗略估计尺度，再用约束项精确求解尺度，并利用反向一致性检验，进一步提高尺度估计的准确性。该文选取了5段典型视频进行实验，并与相关算法进行比较。结果表明所提算法能够消除环境光照、阴影、相似背景和尺度变化等因素所带来的影响，在中心位置误差和跟踪成功率性能指标上，优于其它算法。

Scale-adaptive Object Tracking Based on Color Names Histogram

Tracking effects of algorithms using color information are easily interfered by background clustering, illumination and scale changes, which can result in tracking failure. To solve these problems, an efficient model is proposed to project original RGB color space to a more robust color spaceColor Names (CN) feature space. Furthermore, objects are represented by background weighted color names histogram, and thus the similar background patches around the target are suppressed. Moreover, a two-step tuning way is adapted to estimate the scale by coarse tuning with gradient ascent and fine tuning with constrained items. Back-forward scale check is also used to ensure the precision of scale estimation. 5 representative videos are chosen to examine the proposed algorithms with four others. The results show that the proposed approach is robust to illumination variation, shadows, background clustering, and scale changes. The central distance error and tracking accuracy of the proposed approach also outperform the contrast algorithms.