Please wait a minute...
浙江大学学报(工学版)  2019, Vol. 53 Issue (9): 1741-1748    DOI: 10.3785/j.issn.1008-973X.2019.09.013
计算机科学与人工智能     
基于长方形点过程的遥感图像汽车提取
余煇(),柴登峰*()
浙江大学 空间信息技术研究所,浙江 杭州 310058
Vehicle extraction from remotely sensed images based on rectangle marked point processes
Hui YU(),Deng-feng CHAI*()
Institute of Space Information and Technique, Zhejiang University, Hangzhou 310058, China
 全文: PDF(1963 KB)   HTML
摘要:

针对遥感图像中汽车的识别和提取问题,分析大规模数量汽车的分布和排列,提出基于长方形点过程的汽车分布概率模型. 采用先验模型刻画汽车在空间中的分布规律,采用数据项表达模型与图像的联系. 采用模拟退火策略结合可逆跳马尔科夫链蒙特卡洛采样方法,对模型进行优化求解,实现图像中汽车的自动识别和提取. 实验结果表明,所提方法能很好地用于道路和停车场等规则场景,提取精确度达到99%,回收率达到90%;由于先验模型的约束可以很好地解决汽车相互重叠和方向摇摆的问题,汽车提取的效果比传统算法更好.

关键词: 长方形点过程汽车提取模板匹配模拟退火可逆跳马尔科夫链蒙特卡洛采样    
Abstract:

A rectangle marked point processes was proposed for vehicle identification and extraction from remotely sensed images, and to analyze the distribution and arrangement of a large number of vehicles. A prior model was developed to represent spatial distribution of vehicles, and a data term was applied to establish the relationship between the model and image.The optimal configuration was searched by simulated annealing coupled with Reversible Jump Markov Chain Monte Carlo, so that vehicles could be identified and extracted automatically. The experimental results show that the proposed method works well for the scenes of road and parking lot. The extraction precision is 99%, and the recall is about 90%. The constrain of prior model can solve the problem of vehicle overlapping and uncertain direction well, and the extraction results are better than that of the traditional algorithm.

Key words: rectangle marked point processes    vehicle extraction    template match    simulated annealing    Reversible Jump Markov Chain Monte Carlo
收稿日期: 2018-07-21 出版日期: 2019-09-12
CLC:  TP 751.1  
通讯作者: 柴登峰     E-mail: 21438031@zju.edu.cn;chaidf@zju.edu.cn
作者简介: 余煇(1992—),男,硕士生,从事遥感图像提取算法研究. orcid.org/0000-0002-6600-4739. E-mail: 21438031@zju.edu.cn
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章  
余煇
柴登峰

引用本文:

余煇,柴登峰. 基于长方形点过程的遥感图像汽车提取[J]. 浙江大学学报(工学版), 2019, 53(9): 1741-1748.

Hui YU,Deng-feng CHAI. Vehicle extraction from remotely sensed images based on rectangle marked point processes. Journal of ZheJiang University (Engineering Science), 2019, 53(9): 1741-1748.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2019.09.013        http://www.zjujournals.com/eng/CN/Y2019/V53/I9/1741

图 1  图像中汽车的几何模型
图 2  汽车分布的重叠关系
图 3  汽车分布的方向关系
实验序号 λF α β
1 100 1.0 1.0
2 100 0 1.0
3 1 000 0 1.0
4 1 000 0 0.9
表 1  先验模型中的不同参数设置
图 4  不同参数下的汽车分布结果
图 5  模板匹配处理的图像结果
图 6  道路上汽车图像的提取结果
图 7  广场上汽车图像的提取结果
图像编号 NTP NFP NFN PN RN
1 112 1 8 0.991 2 0.933 3
2 99 1 3 0.990 0 0.970 6
3 802 4 71 0.995 0 0.918 7
4 323 1 18 0.996 9 0.947 2
5 178 1 2 0.994 4 0.988 9
表 2  汽车提取的评估结果
图 8  模板匹配和标点过程的实验结果对比
1 MOON H, CHELLAPPA R, ROSENFELD A Performance analysis of a simple vehicle detection algorithm[J]. Image and Vision Computing, 2002, 20 (1): 1- 13
doi: 10.1016/S0262-8856(01)00059-2
2 TAN Q, WANG J, ALDRED D A. Road vehicle detection and classification from very-high-resolution color digital orthoimagery based on object-oriented method [C] // IEEE International Geoscience and Remote Sensing Symposium. IGARSS 2008. Boston: IEEE, 2008, 4: IV-475-IV-478.
3 GLEASON J, NEFIAN A V, BOUYSSOUNOUSSE X, et al. Vehicle detection from aerial imagery [C] // 2011 IEEE International Conference on Robotics and Automation (ICRA). Shanghai: IEEE, 2011: 2065-2070.
4 LIU K, MATTYUS G Fast multiclass vehicle detection on aerial images[J]. IEEE Geoscience and Remote Sensing Letters, 2015, 12 (9): 1938- 1942
doi: 10.1109/LGRS.2015.2439517
5 CHEN Z, WANG C, LUO H, et al Vehicle detection in high-resolution aerial images based on fast sparse representation classification and multiorder feature[J]. IEEE Transactions on Intelligent Transportation Systems, 2016, 17 (8): 2296- 2309
doi: 10.1109/TITS.2016.2517826
6 CAO L, LUO F, CHEN L, et al Weakly supervised vehicle detection in satellite images via multi-instance discriminative learning[J]. Pattern Recognition, 2017, 64: 417- 424
doi: 10.1016/j.patcog.2016.10.033
7 YANG M Y, LIAO W, LI X, et al. Deep Learning for vehicle detection in aerial images [C] // 2018 25th IEEE International Conference on Image Processing (ICIP). Athens: IEEE, 2018: 3079-3083.
8 STOICA R, DESCOMBES X, ZERUBIA J A Gibbs point process for road extraction from remotely sensed images[J]. International Journal of Computer Vision, 2004, 57 (2): 121- 136
doi: 10.1023/B:VISI.0000013086.45688.5d
9 GE W, COLLINS R T. Marked point processes for crowd counting [C] // IEEE Conference on Computer Vision and Pattern Recognition. Miami: IEEE, 2009: 2913-2920.
10 YU Y, LI J, GUAN H, et al A marked point process for automated building detection from lidar point-clouds[J]. Remote Sensing Letters, 2013, 4 (11): 1127- 1136
doi: 10.1080/2150704X.2013.846487
11 KOWAL M, KORBICZ J. Marked point process for nuclei detection in breast cancer microscopic images [C] // Polish Conference on Biocybernetics and Biomedical Engineering. Poland: Springer, 2017: 230-241.
12 VAN LIESHOUT M N M. Markov point processes and their applications [M]. London: Imperial College Press, 2000.
13 张东萍. 改进 Grabcut 算法在无人机影像船只识别中的应用与研究[D]. 天津: 天津大学, 2013.
ZHANG Dong-ping. Applications and research of improved Grabcut algorithm on ship detection from UAV images[D]. Tianjin: Tianjin University, 2013.
14 CHAI D, SCHMIDT A, HEIPKE C Detecting linear features by spatial point processes[J]. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences-ISPRS Archives, 2016, 41: 841- 848
15 GREEN P J Reversible jump Markov chain Monte Carlo computation and Bayesian model determination[J]. Biometrika, 1995, 82 (4): 711- 732
doi: 10.1093/biomet/82.4.711
16 GAMERMAN D, LOPES H F. Markov chain Monte Carlo: stochastic simulation for Bayesian inference [M]. New York: Chapman and Hall/CRC, 2006.
17 GELMAN A, STERN H S, CARLIN J B, et al. Bayesian data analysis [M]. New York: Chapman and Hall/CRC, 2013.
18 GREEN P J, ?ATUSZY?SKI K, PEREYRA M, et al Bayesian computation: a summary of the current state, and samples backwards and forwards[J]. Statistics and Computing, 2015, 25 (4): 835- 862
doi: 10.1007/s11222-015-9574-5
19 石利平 模拟退火算法及改进研究[J]. 信息技术, 2013, (2): 176- 178
SHI Li-ping Research on simulated annealing algorithm and improvements[J]. Information Technology, 2013, (2): 176- 178
doi: 10.3969/j.issn.1009-2552.2013.02.051
[1] 季琳琳,王清威,周豪,郑美妹. 考虑顾客满意度的冷链水果路径优化[J]. 浙江大学学报(工学版), 2021, 55(2): 307-317.
[2] 黄家骏, 腾来, 张朝杰, 王春晖, 朴成勇. 基于模拟退火算法的I/Q不平衡校正[J]. 浙江大学学报(工学版), 2018, 52(11): 2218-2225.
[3] 袁友伟, 余佳, 郑宏升, 王娇娇. 基于新颖性排名和多服务质量的云工作流调度算法[J]. 浙江大学学报(工学版), 2017, 51(6): 1190-1196.
[4] 程准,鲁植雄,龚佳慧,刁秀永 . 转向系统传递函数的研究及理想传动比获取[J]. 浙江大学学报(工学版), 2016, 50(7): 1276-1283.
[5] 程华强,罗尧治,许贤. 自适应张弦梁结构的非线性内力控制[J]. 浙江大学学报(工学版), 2014, 48(7): 1155-1161.
[6] 叶肖伟,张小明,倪一清,黄启远,樊可清. 基于机器视觉技术的桥梁挠度测试方法[J]. 浙江大学学报(工学版), 2014, 48(5): 813-819.
[7] 张鼎逆,刘毅. 基于改进遗传算法和序列二次规划的再入轨迹优化[J]. J4, 2014, 48(1): 161-167.
[8] 刘业峰,徐冠群,潘全科,柴天佑. 磁性材料成型烧结生产调度优化方法及应用[J]. J4, 2013, 47(9): 1517-1523.
[9] 谢裕江,刘高. 小波边缘分析与建模的波阻抗反演算法的改进——以中国MOU气田盒8段储层分布预测为例[J]. J4, 2013, 47(9): 1680-1684.
[10] 刘爱军, 杨育, 李斐, 邢青松, 陆惠, 张煜东. 混沌模拟退火粒子群优化算法研究及应用[J]. J4, 2013, 47(10): 1722-1730.
[11] 涂岳文, 陈杭, 付秀泉, 李顶立, 黄超, 汤亚伟, 叶树明. 基于心搏聚类的Holter运动伪差段快速识别算法[J]. J4, 2012, 46(6): 1148-1156.
[12] 舒宇, 余锋, 汪乐宇. 基于预留带宽的光纤通道交换网硬实时通信[J]. J4, 2010, 44(9): 1698-1704.
[13] 陈晓刚, 于浩, 王波, 蒋正威, 江全元, 郭创新, 曹一家. 满足多种约束条件的最优PMU配置方法[J]. J4, 2010, 44(3): 539-543.
[14] 林兰芬, 欧冠男, 等. 多约束条件下自动配棉的混合遗传算法[J]. J4, 2009, 43(5): 801-806.