一种基于三维视觉的金属铸锭氧化渣区域感知方法

An Area Perception Method for Oxidized Slag of Metal Ingot Based on 3D Vision

文章提出一种基于三维视觉的金属铸锭氧化渣区域感知方法。首先,针对B-SHOT二进制描述符存在信息丢失、编码错误等问题,提出了改进的WB-SHOT二进制描述符,该描述符增加了新的编码规则,解决了局部编码错误的问题,实现了高准度的氧化渣快速空间特征描述;其次,针对氧化渣点云配准存在精度不高的问题,融合RANSAC粗配准算法和ICP精配准算法,提出一种基于WB-SHOT二进制描述符的粗细两阶段点云配准算法,对空间位置差异较大的模具点云数据和铸锭氧化渣点云数据进行高效配准;最后,结合相似性度量的氧化渣区域分割策略,实现氧化渣三维点云区域的感知,实验结果验证所提方法的有效性。

This paper presents a method of region perception of oxidized slag of metal ingot based on 3D vision.Firstly,aiming at the problems of B-SHOT binary descriptor such as information loss and coding errors,an improved WB-SHOT binary descriptor is proposed,which added a new coding rule,solved the problem of local coding errors,and realized the fast spatial characterization of oxidation slag with high accuracy.Secondly,in order to solve the problem of low accuracy of oxidized slag point cloud registration,a coarse and fine two-stage point cloud registration algorithm based on WB-SHOT binary descriptor is proposed by combining RANSAC coarse registration algorithm and ICP fine registration algorithm,which could efficently register the mold point cloud data and the oxidized slag point cloud data of ingot with large spatial differentces.Finally,the perception of three-dimensional point cloud region of oxidized slag is realized by combining with the region segmentation streategy of similarity measurement,and the effectiveness of the proposed method is verified by experimental results.