您的位置:山东大学 -> 科技期刊社 -> 《山东大学学报(工学版)》

山东大学学报(工学版) ›› 2018, Vol. 48 ›› Issue (2): 1-7.doi: 10.6040/j.issn.1672-3961.0.2016.377

• •    下一篇

基于模板匹配的改进型目标识别算法

丁筱玲1,2,3,赵强1,4,李贻斌1*,马昕1   

  1. 1. 山东大学控制科学与工程学院, 山东 济南 250061;2. 山东农业大学机械与电子工程学院, 山东 泰安 271018;3. 山东省园艺机械与装备重点实验室, 山东 泰安 271018;4. 山东电力设备有限公司, 山东 济南 250022
  • 收稿日期:2016-10-21 出版日期:2018-04-20 发布日期:2016-10-21
  • 通讯作者: 李贻斌(1960— ),男,山东聊城人,教授,博导,主要研究方向为智能机器人技术,智能控制系统等. E-mail:liyb@sdu.edu.cn E-mail:xlding103@163.com
  • 作者简介:丁筱玲(1965— ),女,山东青岛人,教授,博士,主要研究方向为模式识别与智能系统,电子技术应用等. E-mail:xlding103@163.com
  • 基金资助:
    山东省自然基金资助项目(ZR2012CM040);山东省重点研发资助项目(2015GGB01311);山东农业大学现代农业智能化装备研发资助项目(SA24137)

Modified target recognition algorithm based on template matching

DING Xiaoling1,2,3, ZHAO Qiang1,4, LI Yibin1*, MA Xin1   

  1. DING Xiaoling1, 2, 3, ZHAO Qiang1, 4, LI Yibin1*, MA Xin1(1. School of Control Science and Engineering, Shandong University, Jinan 250061, Shandong, China;
    2. Mechanical &
    Electronic Engineering College, Shandong Agricultural University, Tai'an 271018, Shandong, China;
    3. Shandong Provincial Key Laboratory of Horticultural Machineries and Equipments, Tai'an 271018, Shandong, China;
    4. Shandong Power Equipment Company, LTD, Jinan 250022, Shandong, China
  • Received:2016-10-21 Online:2018-04-20 Published:2016-10-21

摘要: 针对图像处理过程中采用局部特征提取与特征匹配的目标识别算法对纹理不丰富物体识别精度差、在同一次学习过程中不能多视角识别同一个物体的缺点,提出采用基于模板匹配的改进型目标识别算法,提高对纹理不丰富物体的识别速度及准确率。利用梯度作为特征量完成模板匹配,结合DOT算法去除次要的梯度特征,只采用幅值大的主导梯度方向作为特征量进行模板匹配,融入仿射投影变换算法、将模板特征二进制化来提高在线同时识别多个不同物体、多视角识别同一个三维物体的速度及准确率。试验证明,该目标识别算法对复杂背景中纹理较少的物体,发生微小变形、微小平移和光照变换的物体识别效果鲁棒性强。

关键词: 主导梯度方向, 目标识别算法, 模板特征二进制化, 模板匹配, 仿射投影变换

Abstract: Because the local feature extraction algorithm got low recognition rate for poor texture object and it could not recognize the same object from different perspective. A modified target recognition algorithm was proposed. The algorithm used gradient characteristics as features to complete the template matching, which only used the main direction of the gradient and the minor gradient feature was removed by DOT(dominant orientation templates)algorithm. The template feature, which was made binary by fusing affine projection transformation algorithm, could improve the recognition rate of identifying multiple objects at the same time or the same object from different perspective. The experiments demonstrated that the proposed algorithm could get better recognition rate and was robust for the object with poor texture, small deformation, small translation and light transformation.

Key words: template characteristics of binary, affine projection transformation, template matching, the dominate gradient direction, target recognition algorithms

中图分类号: 

  • TP391.4
[1] 梅雪,张继法,许松松,等.分级特征提取与选择及在自动目标识别系统中的应用[J].遥感技术与应用,2012,27(5): 712-715. MEI Xue, ZHANG Jifa, XU Songsong, et al. Hierarchical feature extraction and selection method and the applications in automatic target recognition system[J]. Remote Sensing Technology and Application, 2012, 27(5):712-715.
[2] 孔金生, 张小凤,王璇. 基于轮廓特征的模板匹配方法及其应用[J].计算机工程与应用, 2008, 44(22):201-203. KONG Jinsheng, ZHANG Xiaofeng, WANG Xuan. Template matching algorithm based on contour-feature and its application[J]. Computer Engineering and Applications, 2008, 44(22):201-203.
[3] 周晨卉, 王生进, 丁晓青. 基于局部特征级联分类器和模板匹配的行人检测[J]. 中国图象图形学报, 2010, 15(5):824-829. ZHOU Chenhui, WANG Shengjin, DING Xiaoqing. Pedestrian detection based on partial feature and model matching[J].Journal of Image and Graphics, 2010, 15(5):824-829.
[4] 刘忠红,储珺. 特征提取与模板匹配结合的图像拼接方法[J].微计算机信息,2010,26(1):117-118. LIU Zhonghong, CHU Jun. A method for image mosaic based on combining feature extraction with template matching[J].Control & Automation, 2010, 26(1):117-118.
[5] STEGER C. Occlusion, clutter, and illumination invariant object recognition[J]. B Radig & S, 2002, 34:148-154.
[6] 李志军,刘松林,牛照东,等. 基于梯度相位和显著性约束的Hausdorff距离模板匹配方法[J]. 红外与激光工程, 2015, 44(2):775-780. LI Zhijun, LIU Songlin, NIU Zhaodong, et al. Hausdorff distance template matching method based on gradient phase and significance constraints[J].Infrared and Laser Engineering, 2015, 44(2): 775-780.
[7] 肖传民,史泽林,刘云鹏.引入梯度分布特征的图像背景杂波度量[J].光学精密工程,2015,23(12): 3472-3479. XIAO Chuanmin, SHI Zelin, LIU Yunpeng. Metrics of image background clutter by introducing gradient features[J]. Optics and Precision Engineering, 2015, 23(12): 3472-3479.
[8] GRABNER M, GRABNER H, BISCHOF H. Tracking via discriminative online learning of local features[C] //Proceedings of the 2007 IEEE Computer Vision and Pattern Recognition. Minneapolis, USA:IEEE, 2007.
[9] HINTERSTOISSER S, LEPETIT V, ILIC S, et al. Dominant orientation templates for real-time detection of texture-less objects[C] // Proceedings of the 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.San Francisco, USA:IEEE, 2010:2257-2264.
[10] 王国刚,史洪岩,汪滢,等. 仿射不变子空间特征及其在图像匹配中的应用[J]. 红外与激光工程,2014,43(2):659-664. WANG Guogang, SHI Hongyan, WANG Ying, et al. Affine invariant subspace feature and its application in image matching[J]. Infrared and Laser Engineering, 2014, 43(2):659-664.
[11] 王红艳, 高尚兵. 基于全局和局部特征融合的显著性提取方法[J]. 数据采集与处理, 2014, 29(5):801-808. WANG Hongyan, GAO Shangbing.Saliency detection based on fusion of global and local features[J]. Journal of Data Acquisition & Processing, 2014, 29(5):801-808.
[12] 邹明明,卢迪. 基于改进模板匹配的车牌字符识别算法实现[J]. 国外电子测量技术,2010,29(1):59-61. ZOU Mingming, LU Di. Recognition algorithm of car license plate characters based on modified template match[J]. Foreign Electronic Measurement Technology, 2010, 29(1):59-61.
[13] DEGUILLAUME F, VOLOSHYNOVSKIY S V, PUN T. Method for the estimation and recovering from general affine transforms in digital watermarking applications[J]. Proceedings of Spie, 2002, 4675(1):313-322.
[14] CEVIKALP H, TRIGGS B. Visual object detection using cascades of binary and one-class classifiers[J].International Journal of Computer Vision, 2012, 123(10):1-16.
[15] FERRARI V, TUYTELAARS T, GOOL L V. Real-time affine region tracking and coplanar grouping[C] //Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.Kauai, USA: IEEE, 2001(2):226-233.
[16] BAY H, ESS A, TUYTELAARS T, et al. Speeded-up robust features[J]. Computer Vision & Image Understanding, 2008, 110(3):404-417.
[17] LOWE D G. Object recognition from local scale-invariant features[C] //Proceedings of the IEEE International Conference on Computer Vision.Kerkyra, Greece:IEEE, 1999(2):1150-1157.
[18] COMANICIU D, RAMESH V, MEER P. Kernel-based object tracking[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2003, 25(5):564-575.
[19] MATTHEWS I, ISHIKAWA T, BAKER S. The template update problem[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2004, 26(6):810-815.
[20] 李博涛,王之琪,王秀彩.分支界定法在特征向量提取中的应用[J]. 山西电子技术, 2011(2):37-38. LI Botao, WANG Zhiqi, WANG Xiucai. The appication of branch and bound algorithm in eigenvector selection[J]. Shanxi Electronic Technology, 2011(2):37-38.
[21] TAYLOR S, DRUMMOND T. Multiple target localization at over 100 FPS[C] // British Machine Vision Conference, BMVC 2009-Proceedings. London, UK:BMVA. 2009:1-11.
[22] LAMPERT C, BLASCHKO M, HOFMANN T. Beyond sliding windows:object localization by efficient subwindow search[C] //Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Anchorage, USA: IEEE, 2008:1-8.
[1] 王艳超,杨立才*,刘澄玉. 基于模板匹配和镜像延拓的两阶段经验模态分解算法[J]. 山东大学学报(工学版), 2012, 42(6): 69-73.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!