摘要
针对类人机器人的特点,设计了一个具有树叶辨识功能的类人机器人系统。类人机器人自带的摄像头作为视觉,视觉系统采用Sobel算子对除噪、滤波后的图像进行边缘提取,将均匀旋转不变特性与原始的LBP算子相融合,提取边缘图像的特征值,结合差分演化算法对极限学习机进行优化,通过训练得到每类树叶所对应的分类器模型参数,构造分类器,实现类人机器人对树叶准确高效的识别。试验显示,仿人机器人此类控制方式适合参加竞技类比赛。
According to the characteristics of humanoid robot,a humanoid robot system of leaf identification was designed.Robot with type camera as vision,vision system using the Sobel operator for image noise removal filter,edge extraction,the uniform rotation invariant properties of LBP operators and original phase fusion,feature extraction of image edge value,combined with the adaptive differential evolution algorithm to optimize the extreme learning machine,parameter,the classifier model each class leaves obtained by training the classifier was constructed,identification of the humanoid robot to the accurate and efficient.Experiments show that this control mode is suitable for humanoid robot in athletic competition.
出处
《安徽农业科学》
CAS
2014年第16期5295-5297,共3页
Journal of Anhui Agricultural Sciences
基金
中央高校基本科研业务费专项(2013CBQ03)
黑龙江省教育厅科学技术研究项目(12513016)
黑龙江省自然科学基金项目(F201347)
哈尔滨市科技创新人才专项资金项目(2013RFQXJ100)
