论文部分内容阅读
针对象棋机器人下棋时要求识别正确、定位精确的要求,提出了一种基于视觉的中国象棋棋子识别定位技术解决方案。通过对校正后图像应用字符分割、Moravec算子提取字符特征和BP网络算法,实现了棋子的正确识别;把所识别棋子的图像坐标转换成实际棋盘坐标,实现了棋子的准确定位。定位试验结果表明:应用该解决方案的象棋机器人系统能正确识别所有棋子,棋子定位误差在1cm范围以内,最大误差百分比为5.468%,最小误差百分比为0.533%,可以较好地满足象棋机器人的定位要求。
According to the request of chess robot to identify correctly and locate accurately, a vision-based Chinese chess pawn recognition and localization technology solution is proposed. By applying character segmentation to the corrected image, Moravec operator extracts the character features and BP network algorithm to realize the correct recognition of the pawns. The image coordinates of the identified pawns are converted into the actual chessboard coordinates to achieve the accurate positioning of the pawns. The positioning test results show that the chess robot system with this solution can correctly identify all the chess pieces. The positioning error of chess pieces is within 1cm, the maximum error percentage is 5.468% and the minimum error percentage is 0.533%, which can better meet the positioning of the chess robot Claim.