摘要
图像目标主要的两个特征是颜色特征和形状特征,为提高跟踪的准确性和鲁棒性,提出融合图像目标颜色和形状的多特征融合跟踪新方法。采用基于HSV空间的空间颜色概率直方图模型,以及灰度变换后的Hu不变矩模型,分别进行实时目标跟踪,然后采用自适应加权方法,完成图像目标最终跟踪位置的确定。经实验测试,该方法对图像目标受到复杂背景干扰,以及颜色变化、尺度变换以及亮度变化等情况都具有很强的鲁棒性,同时增强跟踪效果,提高了跟踪的有效率。
Color and shape are the two main features of an image target.In order to improve tracking accuracy and robustness,a new multi-feature fusion tracking algorithm fusing color and shape features of the image target is put forward.The histogram model of space color probability based on HSV space,and the Hu invariant moment model after the gray-scale transformation are adopted respectively for real-time target tracking.And then an adaptive weighted method is adopted to confirm the final tracking location of the image target.Two experiments demonstrate that the proposed method has very strong robustness under the circumstances of complex background interference and to the change of color,scale and brightness.At the same time,the algorithm improves the tracking results and the effective rate of tracking.
作者
王智军
王建华
WANG Zhi-jun WANG Jian-hua(The College of Computer and Information Engineering, Chifeng University, Chifeng 024000, China Bethune Medical College, Shijiazhuang 050081, China)
出处
《电光与控制》
北大核心
2017年第11期49-52,57,共5页
Electronics Optics & Control
关键词
运动目标跟踪
图像跟踪
特征融合
颜色直方图模型
HU不变矩
相似性度量
背景干扰
moving target tracking
image tracking
feature fusion
color histogram model
Hu invariant moment
similarity measurement
background interference
