摘要
为了有效消除复杂动态背景对运动物体检测的影响,引入了前后景区分能力较强的YUV颜色空间,基于此空间提出了一种新的码书模型,有效地减少了伪目标的检测。新码书模型采用六元素码字、新的码字学习和更新策略,实现较前人九元素码字和八元素码字有更快的训练速度和更低的存储空间。实验结果表明,即使背景存在运动和光照条件发生变化,算法也能更有效地检测运动目标。
A new codebook model is proposed based on YUV color space which distinguishes foreground and background bet- ter to eliminate the effect of complex dynamic background and remove false targets effectively in the moving object detection. A 6-tuple codeword and a new codebook learning and updating scheme are proposed for fast codebook training and small storage in comparison with the traditional 9-tuple and 8-tuple codeword. Experimental results demonstrate the effectiveness of the algo- rithm proposed in this paper, even when there are moving clutters in the background and illumination variations.
出处
《计算机工程与应用》
CSCD
2013年第16期158-161,共4页
Computer Engineering and Applications
基金
陕西省科技计划(No.2011K09-46)
西安市科技计划工业应用技术研发项目(No.CXY1119)
