Research On The Inspection And Location Algorithm Of FLIP CHIP Based On Machine Vision

With the development of placement machine and electronic packaging technology, FLIP CHIP becomes one of the most popular surface mount components. In the placement machine working processes, it is significant to inspect defects and acquire the position so that it can make sure chips can be placed correctly. This paper is aimed to do some research on inspections and positioning system of FLIP CHIP in placement machine vision system deeply. After analyzing the packaging characteristics and overseas and domestic research status, this paper intends to design chip image preprocessing, chip image characteristic extraction, inspection of chip defect and positioning algorithm diagram. The content below demonstrates the details.Through image preprocessing on noise, uneven illumination and high luminance noise spot which may appear during the FLIP CHIP acquisition process, an ideal binary image can be acquired. Firstly, illumination non-uniformity correction algorithm which is based on probability statistics was used to correct the uneven illumination phenomenon of the original chip image in noise circumstance. After that, maximum between class variance method was chosen to separate target area and background area of chip image. Eventually, open operation and close operation in morphological operation would help ease the white and black inference area after binaryzation of Otsu.Marking the area of binary image could help acquire the target contour. After that, FLIP CHIP’s welded ball contour can be selected based on Hu’s invariants. Interference contour was got rid of from the target contour, and then feature extraction method based on gray discontinuity was adopted by operating the chip’s image contour and Canny edge detection image. Feature extraction method based on gray discontinuity was used to operate the chip’s image contour by comparing another method’s effect. In the meanwhile, after the edge point was filtered, welded ball’s edge point which is set of high precision was obtained.The least square circle fitting method is used to accurately represent the position of the welded ball in the chip. And according to the fitting circle’s parameters, the position, numbers and other essential information can be seen. Compared with standard parameters, the incorrect information would be returned to the system. This algorithm can work without drawing template image and central coordinates of welded ball. In the meanwhile, it can also realize the accurate positioning for asymmetric or irregular welded ball distribution chip. Finally, the completed inspection and positioning algorithm were tested from two aspects which are robustness and generality. The results indicated that.the algorithm proposed in this paper can satisfy the accuracy and speed request of placement machine. The algorithm can also inspect every kind of welded ball distribution chips.