Current manual quality control processes in continuous production environments are limited in impact and reliability. An online automatic inspection system is proposed to generate quality-oriented information regarding the welding of frames in window manufacturing. Using novel edge detection techniques and the Hough transform, a novel algorithm is presented to accurately estimate in real-time welding angles and the overall squareness of the window frame. The inspection system presented is integrated in an automated four-point welding machine and provides decision support to quality control operators regarding the conformity of welded frames. The system is validated in a real scenario and the proposed algorithm estimates welding angles and overall frame squareness with an accuracy of 98.38%, indicating an average error less than 0.8 degrees.