Real time inspection of assembled valve keys in cylinder heads using CCD cameras

The paper presents a vision-based approach for real time inspection and recognition of the assembled valve keys in automobile cylinder heads. The vision solution should be flexible enough to adapt with the valve keys of different cylinder head types according to the requirements. The vision investigation includes cameras, infrared illumination, and an image-processing algorithm. The results of the image-processing algorithm show that either it has found the valve keys in its correct assembled position or not. The total number of correctly positioned valve keys that have been passed by the vision station is also calculated. The inspection is done in real time as cylinder heads are continuously moving on the transportation line. Using images obtained from the camera, we first find the position of the center point of the tappet. Then as a first check angular misalignment of the valve shaft is found out and ultimately as a final check diameter of the valve key is found out. The rejected cylinder heads from the vision station were separated afterwards from the main stream of the transportation line. In the statistics obtained, the vision solution has successfully rejected most of the wrongly assembled valve keys. The false rejection rate was also lower than 0.2% of the total number of the valve keys investigated.