Application Study Of Auto Vision System Based On IC Marking Inspection

Auto Vision System has already been widely applied to every working procedureof Semiconductor manufacturing, especially in the field of marking inspection. Atpresent, more and more requirements for the inspection efficiency, precision,accuracy, reducing power of the AVS are proposed in the industry production. Thispaper studies aforementioned problems theoretically, and an improved the visionsystem has been adopted in the practical production and application now. The structure of Auto Vision System is comprehensively summarized in thispaper. The author proposed how to calibrate the image collection system combiningthe research of system calibration theory. And a practical caliber of improving systemrepeatability and reproducibility has also been put forward. The calibration theorywould be of benefit to the related researcher. A method of Auto-Taught-Pattern based on the theory of auto threshold selectionis proposed in this paper which can improve the auto running speed of AVS. Themarking image would be divided and binarized automatically. The pattern is savedand can be called easily in running the pattern matching. The matching efficiency isgreatly enhanced with the improved matching methodology. Developed achievement has been implemented in Motorola Semiconductormanufacturing. The marking inspection efficiency is improved by 25% through theAuto-Taught-Pattern and advanced Pattern Matching method.