[1]郭晓峰,王耀南,周显恩,等.中国象棋机器人棋子定位与识别方法[J].智能系统学报,2018,13(4):517-523.[doi:10.11992/tis.201709020]
GUO Xiaofeng,WANG Yaonan,ZHOU Xianen,et al.Chess-piece localization and recognition method for Chinese chess robot[J].CAAI Transactions on Intelligent Systems,2018,13(4):517-523.[doi:10.11992/tis.201709020]
点击复制
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
13
期数:
2018年第4期
页码:
517-523
栏目:
学术论文—机器感知与模式识别
出版日期:
2018-07-05
- Title:
-
Chess-piece localization and recognition method for Chinese chess robot
- 作者:
-
郭晓峰1,2, 王耀南1,2, 周显恩1,2, 尹阿婷1,2, 赵辉平1,2, 刘磊1,2
-
1. 湖南大学 电气与信息工程学院, 湖南 长沙 410082;
2. 机器人视觉感知与控制技术国家工程实验室, 湖南 长沙 410082
- Author(s):
-
GUO Xiaofeng1,2, WANG Yaonan1,2, ZHOU Xian’en1,2, YIN A’ting1,2, ZHAO Huiping1,2, LIU Lei1,2
-
1. College of Electrical and Information Engineering, Hunan University, Changsha 410082, China;
2. National Engineering Laboratory for Robot Visual Perception and Control Technology, Changsha 410082, China
-
- 关键词:
-
机器视觉; 圆检测; 字符识别; 最小外接圆定位; 旋转差分识别
- Keywords:
-
machine vision; circle detection; character recognition; minimum circumcircle positioning; rotating differential recognition
- 分类号:
-
TP242
- DOI:
-
10.11992/tis.201709020
- 摘要:
-
针对中国象棋机器人系统中棋子定位与识别问题,提出了一种基于最小外接圆二次定位的定位方法和一种旋转差分识别算法。首先,采用Hough圆检测进行粗定位获取棋子区域,并对棋子进行均值二值化处理。随后,对二值化图像进行形态学处理,提取最大面积轮廓,并利用其最小外接圆实现棋子二次精准定位。最后,对二次定位修正后的图像进行旋转差分识别。以直径为15 mm的棋子为测试对象,利用我们研制的象棋机器人采集图像进行测试,结果表明,棋子的定位精度为0.5 mm,平均定位时间为2.6 ms;在保证棋子识别正确率在98%以上的情况下,单个棋子平均全流程运算时间为10 ms,完全满足现有象棋机器人需求。
- Abstract:
-
To improve the localization and recognition of chess pieces by the Chinese chess robot system, in this paper, we propose a positioning method based on the secondary positioning of the minimum circumcircle and the use of a rotating differential recognition algorithm. First, we use the Hough circle detection method to roughly position chess pieces and then subject them to mean-value binarization. Next, we morphologically process the binarized images to extract the maximum area contours and use their minimum circumcircle values to achieve secondary precise positioning of the chess pieces. Lastly, the rotating differential recognition algorithm recognizes the secondary positioning of the corrected images. Using the chess robot we previously developed and selecting chess pieces with a diameter of 15 mm as test objects, we collected and tested the resulting images. The results show that the positioning accuracy of the chess pieces is within 0.5 mm, and the average positioning time is 2.6 ms. To ensure a 95% recognition accuracy of the chess pieces, the mean full-process operation period for a single chess piece is 10 ms, which fully meets the requirements of the current chess robot.
备注/Memo
收稿日期:2017-09-11。
基金项目:国家自然科学基金项目(61733004,61573134,61433016);国家科技支撑计划项目(2015BAF13B00).
作者简介:郭晓峰,男,1993年生,硕士研究生,主要研究方向为模式识别、机器视觉和图像处理;王耀南,男,1957年生,教授,博士生导师,主要研究方向为电动汽车控制、智能控制理论与应用、智能机器人。曾获国家科技进步二等奖、中国发明创业特等奖、省部科技进步一等奖、省部科技进步二等奖。获国家专利12项。发表学术论文360余篇,其中被SCI检索38篇、EI检索109篇,出版学术专著多部;周显恩,男,1987年生,博士研究生,主要研究方向为模式识别、图像实时处理。
通讯作者:郭晓峰.E-mail:guoxiaofeng@hnu.edu.cn.
更新日期/Last Update:
2018-08-25