A new passive visual sensing system and special composed filter technology was developed firstly in this paper. Through the sensor, clearly welding gap image was collected. After regression recover, mean value smoothing, Laplace enhance, most noise was moved and gap edge became more evident. After analyzing the characters around welding gap, special straight line scanning was used. Then the gap edge was considered near the pixel gray value is the minimum on every scanning straight line. Extensively, because the gray value gradient of the gap edge is the local maximum, so the pixels on the two gap edges were found and were divided into two groups, at last, Hough transform was used to fit the two edges, and gap width and welding direction were calculated according to their definitions.
展开▼
