非凸运动辅助低秩稀疏分解目标检测算法

doi:10.3969/j.issn.1001-506X.2020.06.03

系统工程与电子技术 ›› 2020, Vol. 42 ›› Issue (6): 1218-1225.doi: 10.3969/j.issn.1001-506X.2020.06.03

非凸运动辅助低秩稀疏分解目标检测算法

杨真真^1,²(), 乐俊¹(), 杨永鹏³(), 范露¹()

1. 南京邮电大学通信与网络技术国家工程研究中心, 江苏南京 210023
2. 南京邮电大学理学院, 江苏南京 210023
3. 南京信息职业技术学院网络与通信学院, 江苏南京 210023

收稿日期:2019-09-05 出版日期:2020-06-01 发布日期:2020-06-01
作者简介:杨真真(1984-),女,副教授,博士,主要研究方向为低秩稀疏分解、多媒体信息处理。E-mail:yangzz@njupt.edu.cn|乐俊(1995-),男,硕士研究生,主要研究方向为低秩稀疏分解、多媒体信息处理。E-mail:m15270986042@163.com|杨永鹏(1984-),男,讲师,硕士,主要研究方向为低秩稀疏分解、多媒体信息处理。E-mail:yangyp@cjcit.cn|范露(1994-),女,硕士研究生,主要研究方向为低秩稀疏分解、多媒体信息处理。E-mail:1317014321@njupt.edu.cn
基金资助:
国家自然科学基金(61501251);国家自然科学基金(11671004);中国博士后科学基金(2018M632326);江苏省高校自然科学面上项目(19KJB510044);通信与网络技术国家工程研究中心开放课题(TXKY17010);江苏省高等学校大学生创新创业训练计划项目(创新类项目)(201913112012Y)

Object detection algorithm of nonconvex motion-assisted low rank and sparse decomposition

Zhenzhen YANG^1,²(), Jun LE¹(), Yongpeng YANG³(), Lu FAN¹()

1. National Engineering Research Center of Communications and Networking, Nanjing University of Posts and Telecommunications, Nanjing 210003, China
2. School of Science, Nanjing University of Posts and Telecommunications, Nanjing 210023, China
3. School of Network and Communication, Nanjing Vocational College of Information Technology, Nanjing 210023, China

Received:2019-09-05 Online:2020-06-01 Published:2020-06-01
Supported by:
国家自然科学基金(61501251);国家自然科学基金(11671004);中国博士后科学基金(2018M632326);江苏省高校自然科学面上项目(19KJB510044);通信与网络技术国家工程研究中心开放课题(TXKY17010);江苏省高等学校大学生创新创业训练计划项目(创新类项目)(201913112012Y)

摘要/Abstract

摘要：

针对传统低秩稀疏分解(low rank and sparse decomposition, LRSD)用于视频运动目标检测时检测精度较低的问题,提出了一种鲁棒非凸运动辅助LRSD(robust nonconvex motion-assisted LRSD, RNMALRSD)的运动目标检测算法。该算法首先考虑到视频背景的低秩特性,采用非凸γ范数对秩函数进行逼近,考虑视频背景在变换域上仍然具有稀疏性,引入背景在变换域的稀疏先验。其次,引入运动辅助信息矩阵,使其融入前景的运动信息,表示每个像素属于背景的可能性,提高视频运动目标检测的准确度。然后,采用交替方向乘子法(alternating direction method of multipliers, ADMM)对提出的模型进行求解。最后,将提出的方法应用到视频运动目标检测上进行仿真实验。对实验结果的分析表明，提出的RNMALRSD方法比其他基于LRSD的运动目标检测方法具有更高的检测精度。

关键词: 低秩稀疏分解, 运动辅助, 交替方向乘子法, 鲁棒主成分分析, 目标检测

Abstract:

Aiming at the problem of low detection accuracy when using the traditional low rank and sparse decomposition (LRSD) algorithm for video moving object detection, this paper proposes a moving object detection algorithm based on the robust nonconvex motion-assisted LRSD (RNMALRSD). Firstly, the proposed algorithm considers the low rank characteristic of the background, and utilities the nonconvex γ norm to approximate the rank function. It also considers that the background is still sparse in the transform domain, and introduces the sparse prior of the background in its transform domain. In addition, the motion-assisted information matrix is introduced into the foreground motion information to show the possibility that each pixel belongs to the background and improve the accuracy of video moving object detection. Then, the proposed model is solved by the alternating direction method of multipliers (ADMM). Finally, the proposed method is applied to moving object detection. The experimental results show that the proposed RNMALRSD method has a higher detection accuracy than other moving object detection methods based on LRSD.

Key words: low rank and sparse decomposition (LRSD), motion-assisted, alternating direction method of multipliers (ADMM), robust principal component analysis (RPCA), object detection

中图分类号:

TN912.3

杨真真, 乐俊, 杨永鹏, 范露. 非凸运动辅助低秩稀疏分解目标检测算法[J]. 系统工程与电子技术, 2020, 42(6): 1218-1225.

Zhenzhen YANG, Jun LE, Yongpeng YANG, Lu FAN. Object detection algorithm of nonconvex motion-assisted low rank and sparse decomposition[J]. Systems Engineering and Electronics, 2020, 42(6): 1218-1225.

图/表 3

图1

表1

表2

参考文献 30

1	CANDES E J, LI X D, MA Y, et al. Robust principal component analysis?[J]//Journal of the ACM, 2011, 58(3): 11: 1-11: 37.
2	MU Y D, DONG J, YUAN X T, et al. Accelerated low-rank visual recovery by random projection[C]//Proc.of the Conference on Computer Vision and Pattern Recognition, 2011: 2609-2616.
3	ZHANG H M , QIAN J J , GAO J B , et al. Scalable proximal Jacobian iteration method with global convergence analysis for nonconvex unconstrained composite optimizations[J]. IEEE Trans.on Neural Networks and Learning Systems, 2019, 30 (9): 2825- 2839. doi: 10.1109/TNNLS.2018.2885699
4	KANG Z, PENG C, CHENG Q. Robust PCA via nonconvex rank approximation[C]//Proc.of the IEEE International Confe-rence on Data Mining, 2015: 211-220.
5	XIE W B , YIN H , WANG M N , et al. Low-rank structured sparse representation and reduced dictionary learning based abnormity detection[J]. IET Computer Vision, 2019, 13 (1): 8- 14. doi: 10.1049/iet-cvi.2018.5256
6	FEI L K , XU Y , FANG X Z , et al. Low rank representation with adaptive distance penalty for semi-supervised subspace classification[J]. Pattern Recognition, 2017, 67, 252- 262. doi: 10.1016/j.patcog.2017.02.017
7	WRIGHT J, GANESH A, RAO S, et al. Robust principal component analysis: exact recovery of corrupted low-rank matrices[R]. Urbana-Champaign: Coordinated Science Laboratory, University of Illinois at Urbana-Champaign, 2009.
8	ZHANG X, WANG L X, GU Q Q. A unified framework for nonconvex low-rank plus sparse matrix recovery[C]//Proc.of the 21st International Conference on Artificial Intelligence and Statistics, 2018: 1097-1107.
9	BOUWMANS T , SOBRAL A , JAVED S , et al. Decomposition into low-rank plus additive matrices for background/foreground separation: a review for a comparative evaluation with a large-scale dataset[J]. Computer Science Review, 2017, 23, 1- 71. doi: 10.1016/j.cosrev.2016.11.001
10	LIU X , ZHAO G Y , YAO J W , et al. Background subtraction based on low rank and structured sparse decomposition[J]. IEEE Trans.on Image Processing, 2015, 24 (8): 2502- 2514. doi: 10.1109/TIP.2015.2419084
11	HORNG H L , JEN H C , TYNG L L . Regularized background adaptation: a novel learning rate control scheme for Gaussian mixture modeling[J]. IEEE Trans.on Image Processing, 2011, 20 (3): 822- 836. doi: 10.1109/TIP.2010.2075938
12	BARNICH O , DROOGENBROECK M V . VIBE: a universal background subtraction algorithm for video sequences[J]. IEEE Trans.on Image Processing, 2011, 20 (6): 1709- 1724. doi: 10.1109/TIP.2010.2101613
13	CANDES E, LI X D, MA Y, et al. Robust principal component analysis?: recovering low-rank matrices from sparse errors[C]//Proc.of the IEEE Sensor Array and Multichannel Signal Processing Workshop, 2010: 201-204.
14	ZHOU T Y, TAO D C. GoDec: randomized low-rank & sparse matrix decomposition in noisy case[C]//Proc.of the 28th International Conference on Machine Learning, 2011: 33-40.
15	ZHOU T Y, TAO D C. Bilateral random projections[C]//Proc.of the International Symposium on Information Theory, 2012: 1286-1290.
16	KANG Z, PENG C, CHENG Q. Robust PCA via nonconvex rank approximation[C]//Proc.of the IEEE International Conference on Data Mining, 2015: 211-220.
17	YANG Z Z , YANG Z , HAN D R . Alternating direction method of multipliers for sparse and low-rank decomposition based on nonconvex nonsmooth weighted nuclear norm[J]. IEEE Access, 2018, 6, 56945- 56953. doi: 10.1109/ACCESS.2018.2872688
18	LU C Y , TANG J H , YAN S C , et al. Nonconvex nonsmooth low rank minimization via iteratively reweighted nuclear norm[J]. IEEE Trans.on Image Processing, 2016, 25 (2): 829- 839. doi: 10.1109/TIP.2015.2511584
19	YANG L , PONG T K , CHEN X J . Alternating direction method of multipliers for a class of nonconvex and nonsmooth problems with applications to background/foreground extraction[J]. SIAM Journal on Imaging Sciences, 2017, 10 (1): 74- 110. doi: 10.1137/15M1027528
20	CHEN W G , LI Y L . Stable recovery of low-rank matrix via nonconvex Schatten p-minimization[J]. Science China Mathematics, 2015, 58 (12): 2643- 2654. doi: 10.1007/s11425-015-5081-6
21	XU F , HAN J Q , WANG Y L , et al. Dynamic magnetic resonance imaging via nonconvex low-rank matrix approximation[J]. IEEE Access, 2017, 5, 1958- 1966. doi: 10.1109/ACCESS.2017.2657645
22	GUYON C, BOUWMANS T, ZAHZAH E H. Foreground detection based on low-rank and block-sparse matrix decomposition[C]//Proc.of the 19th IEEE International Conference on Image Processing, 2012: 1225-1228.
23	YE X C , YANG J Y , SUN X , et al. Foreground-background separation from video clips via motion-assisted matrix restoration[J]. IEEE Trans.on Circuits and Systems for Video Technology, 2015, 25 (11): 1721- 1734. doi: 10.1109/TCSVT.2015.2392491
24	XUE Z C , DONG J , ZHAO Y X , et al. Low-rank and sparse matrix decomposition via the truncated nuclear norm and a sparse regularizer[J]. The Visual Computer, 2019, 35, 1549- 1566. doi: 10.1007/s00371-018-1555-1
25	QIAN W , CAO F . Adaptive algorithms for low-rank and sparse matrix recovery with truncated nuclear norm[J]. International Journal of Machine Learning and Cybernetics, 2019, 10, 1341- 1355. doi: 10.1007/s13042-018-0814-9
26	PORWIK P , LISOWSKA A . The Haar-wavelet transform in digital image processing: its status and achievements[J]. Machine Graphics and Vision, 2004, 13 (1/2): 79- 98.
27	TAO P D , AN L T H . Convex analysis approach to D.C.programming: theory, algorithms and applications[J]. Acta Mathematica Vietnamica, 1997, 22 (1): 287- 367.
28	LI L Y , HUANG W M , GU Y H , et al. Statistical modeling of complex backgrounds for foreground object detection[J]. IEEE Trans.on Medical Imaging, 2004, 13 (11): 1459- 1472.
29	GOYETTE N, JODOIN P M, PORIKLI F, et al. Changedetection.net: a new change detection benchmark dataset[C]//Proc.of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, 2012. DOI: 10.1109/CVPRW.2012.6238919.
30	BRUTZER S, HOFERLIN B, HEIDEMANN G. Evaluation of background subtraction techniques for video surveillance[C]//Proc.of the Conference on Computer Vision and Pattern Recognition, 2011: 1937-1944.

算法	视频帧
算法	Bootstrap	ShoppingMall	Office	WaterSurface	Highway	Backdoor	BusStation	平均值
RNMALRSD	0.742 7	0.739 5	0.836 6	0.812 4	0.733 3	0.612 9	0.776 3	0.750 0
MAMR	0.742 3	0.734 5	0.823 5	0.798 3	0.729 6	0.609 7	0.772 7	0.744 4
LRSD-TNNSR	0.666 0	0.732 7	0.711 8	0.787 5	0.731 5	0.607 7	0.763 2	0.714 3
GoDec	0.738 7	0.683 0	0.691 1	0.761 4	0.725 5	0.574 7	0.721 0	0.699 3
PCP	0.705 2	0.718 5	0.496 6	0.776 1	0.708 1	0.611 3	0.760 9	0.682 4
LRSD-ATNN	0.658 8	0.719 2	0.733 5	0.806 7	0.713 0	0.602 1	0.754 1	0.712 5
NRPCA	0.679 2	0.718 5	0.739 3	0.785 3	0.731 2	0.609 4	0.772 6	0.719 4

算法	视频帧
算法	Bootstrap	ShoppingMall	Office	WaterSurface	Highway	Backdoor	BusStation	平均值
RNMALRSD	0.738 5	0.729 3	0.759 8	0.807 5	0.732 3	0.611 2	0.741 5	0.731 5
MAMR	0.737 5	0.726 1	0.713 4	0.774 5	0.703 6	0.607 0	0.738 4	0.714 4
LRSD-TNNSR	0.651 8	0.720 8	0.520 5	0.768 4	0.703 3	0.608 6	0.719 6	0.670 4
GoDec	0.719 7	0.707 4	0.600 3	0.753 6	0.701 5	0.574 0	0.694 9	0.678 8
PCP	0.691 2	0.710 5	0.493 5	0.772 7	0.661 0	0.593 0	0.647 3	0.652 7
LRSD-ATNN	0.651 4	0.719 7	0.510 4	0.794 0	0.702 6	0.599 5	0.718 7	0.670 9
NRPCA	0.682 4	0.726 9	0.640 5	0.765 6	0.700 5	0.606 1	0.687 3	0.687 0

[1]	宋爽, 张悦, 张琳娜, 岑翼刚, 李浥东. 基于深度学习的轻量化目标检测算法[J]. 系统工程与电子技术, 2022, 44(9): 2716-2725.
[2]	肖宇, 邓正宏, 张展. 基于双阶段互信息准则的多目标检测波形设计[J]. 系统工程与电子技术, 2022, 44(9): 2736-2742.
[3]	刘祥, 黄天耀, 刘一民. 频率捷变雷达的扩展目标检测[J]. 系统工程与电子技术, 2022, 44(6): 1833-1838.
[4]	赵晓枫, 徐叶斌, 吴飞, 牛家辉, 蔡伟, 张志利. 基于全局感知机制的地面红外目标检测方法[J]. 系统工程与电子技术, 2022, 44(5): 1461-1467.
[5]	魏文晓, 刘洁瑜, 沈强, 李成. 基于人眼视点图的特征融合小目标检测算法[J]. 系统工程与电子技术, 2022, 44(4): 1120-1127.
[6]	陈玉香, 陈胜垚, 冉龙瑶, 席峰. 动态超表面天线阵列混合发射波束成形[J]. 系统工程与电子技术, 2022, 44(3): 730-736.
[7]	李洪瑶, 李小强, 韩心中, 谢学立, 席建祥. 基于决策融合的多无人机协同目标检测识别算法[J]. 系统工程与电子技术, 2022, 44(3): 746-754.
[8]	杨磊, 张苏, 盖明慧, 方澄. 高分辨SAR目标成像方向性结构特征增强[J]. 系统工程与电子技术, 2022, 44(3): 808-818.
[9]	潘超凡, 李润生, 许岩, 胡庆, 牛朝阳, 刘伟. 基于感知向量的光学遥感图像舰船检测[J]. 系统工程与电子技术, 2022, 44(12): 3631-3640.
[10]	贾晓雅, 汪洪桥, 杨亚聃, 崔忠马, 熊斌. 基于YOLO框架的无锚框SAR图像舰船目标检测[J]. 系统工程与电子技术, 2022, 44(12): 3703-3709.
[11]	薛春岭, 曹菲, 孙庆, 秦建强, 冯晓伟. 基于多特征信息融合的海面微弱目标检测[J]. 系统工程与电子技术, 2022, 44(11): 3338-3345.
[12]	李永刚, 朱卫纲, 黄琼男, 李云涛, 何永华. 复杂背景下SAR图像近岸舰船目标检测[J]. 系统工程与电子技术, 2022, 44(10): 3096-3103.
[13]	范加利, 田少兵, 黄葵, 朱兴动. 基于Faster R-CNN的航母舰面多尺度目标检测算法[J]. 系统工程与电子技术, 2022, 44(1): 40-46.
[14]	龚树凤, 龙伟军, 贲德, 潘明海. 组网雷达自适应模糊CFAR检测融合算法[J]. 系统工程与电子技术, 2022, 44(1): 100-107.
[15]	杨威, 刘永祥, 付耀文, 黎湘. 基于连续信任函数理论的鲁棒雷达目标检测方法[J]. 系统工程与电子技术, 2021, 43(9): 2463-2469.

非凸运动辅助低秩稀疏分解目标检测算法

Object detection algorithm of nonconvex motion-assisted low rank and sparse decomposition

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 3

参考文献 30

相关文章 15

编辑推荐

Metrics

本文评价