一种基于TDH的手绘图形方向识别方法

J4 ›› 2013, Vol. 35 ›› Issue (12): 141-145.

一种基于TDH的手绘图形方向识别方法

张友根1，吴玲达1,2，宋汉辰1

(1.国防科学技术大学信息系统工程重点实验室,湖南长沙 410073;2.装备学院重点实验室,北京 101416)

收稿日期:2012-07-04 修回日期:2012-09-11 出版日期:2013-12-25 发布日期:2013-12-25
基金资助:
国家自然科学基金资助项目（61103081）

A TDH-based orientation recognition method for sketched symbols

ZHANG Yougen1,WU Lingda1,2,SONG Hanchen1

(1.Science and Technology on Information Systems Engineering Laboratory,
National University of Defense Technology,Changsha 410073;
2.Key Laboratory,Academy of Equipment,Beijing 101416,China)

Received:2012-07-04 Revised:2012-09-11 Online:2013-12-25 Published:2013-12-25

摘要/Abstract

摘要：

在方向可变的手绘图形识别中，要使用对方向敏感的特征及识别方法，识别图形的方向是一个前提和关键性的问题。提出了一种基于切向直方图TDH特征的手绘图形方向识别方法，利用图形采样点集的局部切方向统计特征TDH描述手绘图形的笔画方向信息，通过TDH旋转匹配确定待识别图形的方向相对于模板图形方向的旋转角度，进而可将二者旋转对齐。在包含7类共52种点军标的手绘图形集上的实验结果表明，基于TDH的方法对超过80%的图形所得的方向识别角度误差在5°以内，识别速度满足交互实时性的要求，验证了本文方法的有效性。

关键词: 手绘图形, 方向识别, 切向直方图

Abstract:

In order to apply rotation sensitive features and recognition methods for recognizing rotatable sketched symbols, it is essential to recognize its orientation. A TDHbased orientation recognition method is proposed for sketched symbols. Tangent directions of resample points are extracted, and a histogram feature is constructed to carry directional information of the symbol’s strokes. The rotation angle between an unknown input symbol and a template symbol can be estimated by matching their TDH rotationally, and the input symbol can be adjusted to align with the template one subsequently. With an experiment on a set of 52 sketched courseofaction symbols, the results show that the error of rotation angle recognized by our method is less than 5° for over 80% of the tested symbols, and the recognition speed meets the real time requirements.

Key words: sketched symbols;orientation recognition;tangent direction histogram(TDH)

张友根1，吴玲达1,2，宋汉辰1. 一种基于TDH的手绘图形方向识别方法[J]. J4, 2013, 35(12): 141-145.

ZHANG Yougen1,WU Lingda1,2,SONG Hanchen1. A TDH-based orientation recognition method for sketched symbols [J]. J4, 2013, 35(12): 141-145.

参考文献

［1］Johnson G, Gross M D, Hong J, et al.Computational support for sketching in design:A review［J］.Foundations and Trends in Human Computer Interaction,2009,2(1):193.
［2］Hse H,Newton A R.Sketched symbol recognition using Zernike moments［C］∥Proc of the 17th International Conference on Pattern Recognition, 2004:367370.
［3］Su Yang.Symbol recognition via statistical integration of pixellevel constraint histograms:A new descriptor［J］.IEEE Transactions on Pattern Analysis and Machine Intelligence,2005,27(2):278281.
［4］Kara L B, Stahovich T F. An imagebased, trainable symbol recognizer for handdrawn sketches［J］.Computers & Graphics,2005,29(4):501517.
［5］Fornés A, Lladós J, Sánchez G, et al. Rotation invariant handdrawn symbol recognition based on a dynamic time warping model［J］.International Journal on Document Analysis and Recognition,2010,13(3):229241.
［6］Ouyang T Y, Davis R. A visual approach to sketched symbol recognition［C］∥Proc of the 21st International Joint Conference on Artifical Intelligence, 2009:14631468.
［7］Leung K C, Leung C H.Recognition of handwritten Chinese characters by critical region analysis［J］.Pattern Recognition,2010,43(3):949961.
［8］Huang Shengming, Jin Lianwen, Lv Jin. A novel approach for rotation free online handwritten Chinese character recognition［C］∥Proc of the 10th International Conference on Document Analysis and Recognition, 2009:11361140.
［9］Reaver J, Stahovich T F, Herold J. How to make a quick$:Using hierarchical clustering to improve the efficiency of the dollar recognizer［C］∥Proc of the 8th Eurographics Symposium on SketchBased Interfaces and Modeling, 2011:103108.
［10］He Tingting,Huo Qiang. A characterstructureguided approach to estimating possible orientations of a rotated isolated online handwritten Chinese character［C］∥Proc of the 10th International Conference on Document Analysis and Recognition, 2009:536540.
［11］Revaud J, Lavoué G, Baskurt A. Improving Zernike moments comparison for optimal similarity and rotation angle retrieval［J］.IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(4):627636.
［12］Rosin P L,Zunic J.Orientation and anisotropy of multicomponent shapes from boundary information［J］.Pattern Recognition,2011,44(9):21472160.