Automatic Pouring System for Metal Casting based on Learning from Examples

PTZ Camera is used to acquire image sequences of metal cast pouring process,and an automatic pouring algorithm is proposed based on learning from examples. This algorithm includes two stages: training from examples and automatic pouring. At the first stage,it takes manual operation of metal cast pouring as examples,and takes multiple frames to estimate pouring position,pouring basin area,and pouring time. In the second stage,it combines pouring basin area and pouring time to control the open/close of sprue plug,and it updates pouring parameters to direct the motion of PTZ camera automatically,thus,to make pouring process lie around the center of PTZ camera views all the time. Experiment results show that this system achieves high robustness and precision,and improves the efficiency of product line of metal casing greatly.