针对运动目标实时检测的改进码书建模方法
首发时间:2010-04-22
摘要:针对码书模型因算法复杂度高而难以应用到运动目标实时检测中的问题,提出了一种改进的码书建模方法。首先优化了码书模型的参数和训练策略,然后提出一种信息块快速获取策略,利用该策略对获得的目标信息块和时间轴运动信息块进行融合决策,实现了在检测过程中实时分离背景和训练码书模型。实验结果表明:经优化的码书建模方法处理效率大幅度提高的同时,检测性能与原始码书模型一致,能对多变环境快速适应,有效解决了码书建模的实时运动目标检测问题。
For information in English, please click here
Codebook improved modeling method for real-time moving target detection
Abstract:Currently, the codebook modeling method is difficult to be applied in real-time video moving target detection for its’ high computational complexity. In order to solve the problem, an improved codebook modeling strategy in real-time application is proposed. First, parameters and training strategy of the codebook are optimized. Then, a rapid information block acquisition strategy is presented, which can be used to get background and train codebook model in real-time with composite decision from the result of target information block and moving information block. The experimental results show that the performance of the proposed method is close to the multi-parameter codebook modeling strategy, and this method is capable of adapting to the variations of outdoor environment. Hence, the method can effectively solve the problem of codebook model applied in real-time environment.
Keywords: moving targets detection codebook model background learning background modeling
论文图表:
引用
No.4221952828712719****
同行评议
共计0人参与
勘误表
针对运动目标实时检测的改进码书建模方法
评论
全部评论0/1000