摘要
在机器人世界杯比赛中,对球员的识别是进行路线规划、传球等上层策略的基础。由于NAO机器人硬件条件的限制和场外环境的干扰,对算法的运算速度和准确度有一定的要求。为满足这些要求,本文提出将类Haar特征和级联Adaboost算法应用到对NAO的识别中。首先,通过在离线环境下由Adaboost算法训练得到的级联分类器对NAO进行初次识别;然后,利用颜色直方图匹配对候选目标区域进行二次识别,在排除误检区域的同时,也进行敌我识别。实验结果表明,本方法能够有效地识别NAO机器人,满足比赛中识别算法对鲁棒性和实时性的要求。
In the Robo Cup games,recognition of the players is the basis of route planning,passing and other top strategy. Due to the limitation of hardware condition of NAO robot and the disturber of the background,the speed and accuracy of the algorithm are required. In order to meet these requirements,a recognition algorithm based on Haar-like features and cascade Adaboost classifier is proposed. Firstly,NAO was first recognized by a cascade classifier trained by the Adaboost algorithm in the offline environment; then,the candidate target regions are identified by using the color histogram matching for second times,in the removal of false detection area at the same time,NAO also identified other robots being friend or foe. The experimental results show that this method can effectively identify the NAO robot,and meet the requirements of the robustness and real-time of the recognition algorithm.
出处
《计算机与现代化》
2017年第2期30-35,共6页
Computer and Modernization
基金
国家自然科学基金资助项目(61272220)
