Experimental Study on Recognation of Bulk Materials Level by SVM Posterior Probability

Abstract

Abstract:

A image-based method for bulk level detection in bulk loading operations was proposed herein. The method first divided the image to several non-overlapping sub-blocks, then each blocks were filtered by a homomorphic filter, converted to binary image, abstracted the binary co-occurrence matrix, and recognized by SVM according to their texture features. In recognition process, a target-proportion model of the blocks which located between the boundary of bulk and the background was found based on SVM posterior probability. The model combined the SVM bulk level fitting results and the SVM posterior probability values was used to segmentation process in these boundary-blocks. Experimental results indicate that the recognition errors of the contour and the bulk level are reduced by 42% and 56% compared to the SVM recognition algorithm, average errors are as 0.4517pixel and 0.2586pixel respectively. The running times of the proposed method is as 0.2s under MATLAB.

Key words: machine vision, bulk loading;level detection, texture recognition, support vector machine(SVM) posterior probability

摘要：

提出了一种散料装车料位的图像检测方法，该方法将待处理料堆图像分为若干不重叠的子块，对各子块进行了同态滤波、二值化及二值共生矩阵纹理特征提取，并根据纹理特征对各子块进行了分类识别。在识别过程中，提出了一种基于SVM及其后验概率的料堆识别方法，建立了位于交界位置子块的SVM后验概率与其中料堆目标所占比例的关系模型，并将仅采用SVM对子块识别后的料位拟合结果与其后验概率输出相结合，在这些交界位置子块内进行了进一步的图像分割。试验结果表明，所提出的方法与仅采用SVM子块识别的料堆轮廓及料位拟合误差相比，分别减小42%和56%，平均误差分别为0.4517pixel和0.2586pixel，在MATLAB下每帧处理需0.2s。

关键词: 机器视觉, 散料装车, 料位检测, 纹理识别, SVM后验概率

CLC Number:

TP391.4

Tian Mingrui, Hu Yongbiao, Jin Shoufeng. Experimental Study on Recognation of Bulk Materials Level by SVM Posterior Probability[J]. China Mechanical Engineering, 2016, 27(05): 646-651.

田明锐, 胡永彪, 金守峰. 结合SVM后验概率的散料料位识别试验研究[J]. 中国机械工程, 2016, 27(05): 646-651.

References

[1]Fletcher E. Rapid Load out Systems Past Present and Future.[2008-02-26]. http://www.ckit.co.za/right-index/ tech-focus/rapid-load/rapid-load.html.

[2]Sexton A, Bennett D. Discuss Automation Trends in TruckLoading[J]. World Coal, 2012,6(1): 96-100.

[3]刘朝荣. 装车系统矿石料位控制方法的研究[J]. 电气自动化, 2012(4): 16-17.

Liu Chaorong. Research on Control Method for Ore material Level of the Ore-loading System[J]. Intelligent Control Techniques, 2012(4): 16-17.

[4]张广军，孙海，张新，等. 一种带有自动布料平车装置的汽车快速定量装车站: 中国,201010126967.6[P]. 2010-08-04.

[5]闫玉刚，孟凡英.自动输送汽车衡: 中国,200910209808.X[P]. 2010-12-22.

[6]王建, 姚振强, 包起帆，等. 新型全自动抓斗卸船系统[J]. 解放军理工大学学报(自然科学版), 2010, 11(3): 284-289.

Wang Jian, Yao Zhenqiang, Bao Qifan, et al. Novel Full-automatic Grab Ship Unloader System[J]. Journal of PLA University of Science and Technology(Nature Science Edition), 2010, 11(3): 284-289.

[7]包起帆. 港口散货全自动装卸设备的研究与开发[J]. 中国机械工程, 2008, 19(23):2797-2803.

Bao Qifan. Research and Development of Automatic Bulk Cargo Equipment in Modern Ports[J]. China Mechanical Engineering, 2008, 19(23):2797-2803.

[8]Tentz A, Bares J, Singha S. Robotic Excavator for Autonomous Truck Loading[J]. Autonomous Robots, 1999,7:175-186.

[9]Riku P. Method and Apparatus for Automatic Loading of Dumper:United States,7751927[P]. 2010-07-06.

[10]黄国鹏,黄国坤. 一种基于图像的露天矿挖掘机装车过程的监控方法和装置:中国,201110095634.6[P]. 2011-11-16.

[11]王福斌, 刘杰, 代沅兴，等. 履带式挖掘机器人CCD道路识别[J]. 中国工程机械学报, 2010,8(2): 230-233.

Wang Fubin, Liu Jie, Dai Ruanxing, et al. CCD-enabled Road Recognition for Caterpillar Excavator Robots[J]. Chinese Journal of Construction Machinery, 2010, 8(2): 230-233.

[12]肖婷, 文怀兴, 夏田. 基于双目识别技术的挖掘机智能化控制系统[J]. 工程机械, 2007, 38(3): 4-7.

Xiao Ting, Wen Huaixing, Xia Tian. Intelligent Control System of Excavator Based on Binocular Stereo-vision[J].Construction Machinery and Equipment,2007,38(3):4-7.

[13]田明锐, 胡永彪, 金守峰,等. 基于图像纹理的散料装车料位识别试验研究[J]. 中国机械工程, 2013, 24(7): 910-914.

Tian Mingrui, Hu Yongbiao, Jin Shoufeng,et al. Experimental Study on Level Detection of Bulk Materials Entrucking Based on Image Texture[J]. China Mechanical Engineering, 2013, 24(7): 910-914.

[14]王芳,江剑,石秀东,等.沥青混凝土成品装车指挥仪的研制[J].微计算机信息,2009,25(29):14-15.

Wang Fang, Jiang Jian, Shi Xiudong, et al. Development of the Director in the Load of Finished Asphalt Concrete[J]. Microcomputer Information, 2009,25(29):14-15.

[15]吴延明,郝晓弘.平硐矿车智能装车系统的研究与实现[J].工矿自动化,2008(4):106-107.

Wu Yanming, Hao Xiaohong. Research and Implementation of the Intelligence Adit Harvester Loading System[J].Industry and Mine Automation, 2008(4):106-107.

[16]Achanta R, Shaji A, Smith K, et al. SLIC Superpixels Compared to State of the art Superpixel Methods[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(11):2274-2281.

[17]Levinshtein A, Stere A, Kiriakos N. TurboPixels: Fast Superpixels Using Geometric Flows[C]//IEEE Trans Pattern Anal. Mach. Intell..Toronto, 2009:2290-7.

[18]He X, Zemel R, Ray D. Learning and Incorporating Top-down Cues in Image Segmentation[C]//ECCV.Graz,2006：338-351.

[19]贺晓辉,鄢萍,刘飞,等. 基于支持向量机和功率信息的工件在线识别方法[J]. 中国机械工程, 2011, 22(7): 818-824.

He Xiaohui, Yan Ping, Liu Fei, et al. Workpiece Online Identification Method Based on SVM and Power Information[J]. China Mechanical Engineering, 2011, 22(7): 818-824.

[20]刘丽,匡纲要. 图像纹理特征提取方法综述[J]. 中国图像图形学报, 2009, 14(4): 622-635.

Liu Li, Kuang Gangyao. Overview of Image Textural Feature Extraction Methods[J]. Journal of Image and Graphics, 2009, 14(4): 622-635.

[21]薄华, 马缚龙, 焦李成，等. 图像纹理的灰度共生矩阵计算问题的分析[J]. 电子学报, 2006, 34(1): 155-158, 134.

Bo Hua, Ma Fulong, Jiao Licheng. Research on Computation of GLCM of Image Texture[J]. Acta Electronic Sinica,2006,34(1):155-158,134.

[22]肖俊,宋寿鹏,丁丽娟. 空域同态滤波算法研究[J]. 中国图像图形学报, 2008,13(12): 2302-2306.

Xiao Jun, Song Soupeng, Ding Lijuan. Research on the Fast Algorithm of Spatial Homomorphic Filtering[J]. Journal of Image and Graphics, 2008,23(12): 2302-2306.

[23]Platt J C. Probabilistic Output for Support Vector Machine and Comparisons to Regularized LikelihoodMethods[M]. Pennsylvania: MIT Press, 1999.