船体小组立特征提取的机器视觉方法与实现

为了实现无先验信息、无人工决策干预的船体小组立智能制造,采用机器视觉技术来实现小组立表面特征提取与定位。根据船体小组立智能生产线自动焊接定位的需求,设计对小组立工件激光扫描点云进行识别与提取的机器视觉方法,从降低复杂度和运算时间的角度对算法进行了优化,并开发了相应软件。通过不同小组立工件特征提取与定位实验,验证了本文算法在无先验信息、无人工干预、工件随机摆放条件下的可靠性,表明本文所介绍的机器视觉方法能够适应船体构件种类多、非标准件比例高、装配位置随机的特点,为船体智能制造实现提供支持。

Methodology and Implementation of Machine Vision for Ship Small Assembly Feature Detection

Feature detection of ship small assembly is prerequisite of ship small assembly intelligent manufacturing without priori knowledge or human-relied input. Methodology of machine vision system for small assembly recognition from point cloud are designed and optimized to reduce its complexity and time cost. A machine vision software is also developed. Experiments of random-positioned small assemblies are also conducted to approve reliability of the machine vision system without priori knowledge or human-relied input. The machine vision system is capable providing input information for small assembly automatic welding. This work will help to tackle problems in ship intelligent manufacturing.