工件形状的激光雷达三维扫描识别方法  

Three-dimensional lidar scanning recognition method of workpiece shape

作　　者：张科研 孙继旋 庞启硕 ZHANG Keyan;SUN Jixuan;PANG Qishuo(Beibu Gulf University,Qinzhou Guangxi 535011,China)

机构地区：[1]北部湾大学,广西钦州535011

出　　处：《激光杂志》2021年第5期133-137,共5页Laser Journal

基　　金：广西高校中青年教师科研基础研究能力提升项目(No.2020KY10015)。

摘　　要：当前激光雷达三维扫描识别方法运算复杂,存在工件形状识别精度较低、识别时间较长的问题。因此,提出激光雷达三维扫描工件形状的识别方法。采用激光雷达三维扫描技术得到目标工件的中心与方向;通过灰度化算法实现RGB图像转换,利用非线性滤波方法完成转换图像的中值滤波;根据边界型和区域型的工件特征参数,结合旋转、平移以及缩放比例尺等不变的特征,运用傅里叶描述子算法完成三维图像特征提取,实现机械工件形状的激光雷达三维扫描识别。实验结果表明,本方法的工件形状识别精度较高,平均识别精度高达98.97%,识别时间仅为15 s。At present,the 3 d scanning recognition method of LIDAR is complicated,with some problems such as low precision and long recognition time. Therefore,the recognition method of 3 d scanning workpiece shape by li DAR is proposed. The centre and direction of the target workpiece are obtained by using laser radar 3 d scanning technology.The greying algorithm realises RGB image conversion,and median filtering is completed by the non-linear filtering method. According to the feature parameters of the boundary and regional workpiece,combining the invariant features of rotation,translation and scale,the Fourier descriptor algorithm is used to complete the feature extraction of 3 d image and realise the LIDAR 3 D scan recognition of the shape of the mechanical workpiece. The experimental results show that the proposed method’s workpiece recognition accuracy is high,with an average recognition accuracy of 98. 97%and a recognition time of only 15 s.

关 键 词：激光雷达 三维扫描 机械工件形状 傅里叶变换 

分 类 号：TN391[电子电信—物理电子学]