张桂梅, 姚伟. 一种基于低秩矩阵的车流量检测算法[J]. 南昌航空大学学报(自然科学版), 2014, 28(4): 60-66. DOI: 10.3969/j.issn.1001-4926.2014.04.012
引用本文: 张桂梅, 姚伟. 一种基于低秩矩阵的车流量检测算法[J]. 南昌航空大学学报(自然科学版), 2014, 28(4): 60-66. DOI: 10.3969/j.issn.1001-4926.2014.04.012
ZHANG Gui-mei, YAO Wei. A Vehicle Flow Detection Algorithm Based on Low-Rank Matrix[J]. Journal of nanchang hangkong university(Natural science edition), 2014, 28(4): 60-66. DOI: 10.3969/j.issn.1001-4926.2014.04.012
Citation: ZHANG Gui-mei, YAO Wei. A Vehicle Flow Detection Algorithm Based on Low-Rank Matrix[J]. Journal of nanchang hangkong university(Natural science edition), 2014, 28(4): 60-66. DOI: 10.3969/j.issn.1001-4926.2014.04.012

一种基于低秩矩阵的车流量检测算法

A Vehicle Flow Detection Algorithm Based on Low-Rank Matrix

  • 摘要: 针对传统车流量检测方法在复杂环境中检测精度较低的问题,提出了一种新的基于低秩矩阵的车流量检测方法。首先利用伊辛模型和鲁棒性主成份分析方法(RPCA)得到非凸的能量函数,然后利用奇异值分解(SVD)并且不断迭代的方法分步解决能量函数非凸性的问题,进而优化能量函数检测出最佳车辆前景,最后利用虚拟检测线圈来统计车流量。实验结果表明:该方法与帧差法和混合高斯算法相比,检测车流量的精度得到显著提高,并且能够较好地分割大雾天气的运动车辆。

     

    Abstract: The traditional detection method of vehicle flow detection have limitations to low accuracy in the complex scene, this paper proposes a new vehicle flow detection algorithm based on low-rank matrix. The algorithm firstly introduce the Ising model and Robust Principal Component Analysis (RPCA) to get the no-convex energy function, and then employ the singular value decomposition (SVD) and iterate step by step to solve the problem that energy function is non-convex, and then optimize the energy function to detect the foreground vehicles. Finally, we count the number of vehicles by using virtual coil. Compared with the frame-difference method and the mixed Gaussian algorithm, the experimental results show that the proposed method can detect vehicle effectively and accurately, even in fog weather.

     

/

返回文章
返回