摘要
由于颜色直方图只考虑了目标颜色的统计信息,忽略了目标颜色的空间信息,使得在模板匹配过程中容易收敛到其他位置,从而导致目标丢失。针对这一缺点,提出了一种基于颜色-空间二维直方图的目标跟踪技术,在建立目标模板的过程中融入了目标颜色的空间信息。实验结果表明,与传统的基于颜色直方图的Mean-shift算法相比,基于颜色-空间二维直方图的目标跟踪技术能更准确的对目标进行跟踪,提高了目标跟踪的鲁棒性。
Because of only concerning about the object color statistics information and ignoring the spatial information, the color histogram can easily convergence to a different location in the process of tracking target and result in the loss of target. Aim at this disadvantage, an object tracking technology based on color-space 2D histogram was proposed in this article, which in- tegrates the object's spatial information in the process of establishing the target model. Experiments demonstrate that compared with the mean-shift algorithm based on color histogram, the proposed method can track the object more exactly and promote the stability of object tracking greatly.
出处
《计算机与数字工程》
2009年第12期126-129,共4页
Computer & Digital Engineering
基金
福建省自然科学基金项目(编号:A0510004)
福州大学科技发展基金项目(编号:2007-XQ-25)资助