Research On Voids Detection Technology In BGA Solder Balls Based On X-ray Image

With the increasing popularity of BGA(Ball Grid Array)packaging technology,the corresponding welding quality detection technology is becoming more and more important.Because of the fact that X-ray inspection technology can effectively detect the internal defects of solder balls,it has become a very important technology in the welding quality inspection of BGA packaging devices.With the rapid development of integrated circuits,the integration of circuit board is becoming higher,and the surrounding environment of BGA is becoming more complex,which makes it more difficult to detect BGA solder ball voids based on traditional methods.Therefore,aiming at the problems of slant image,difficult of the identification and extraction of voids in BGA solder balls based on X-ray detection,a series of researches have been carried out.Firstly,aiming at the problem that the X-ray image of BGA solder balls may be slanted,a method which can automatically detect and correcting the X-ray image of BGA solder balls is proposed.In this method,the rough coordinates of solder balls in the original image are identified by Hough circle detection.The topological relationship of solder balls in the original image is established by Delaunay triangulation algorithm.Then,the coordinates of solder balls in ideal positive situation are estimated to calculate the transformation matrix between the actual coordinates and the ideal coordinates of solder balls.At last,the corrected image is obtained by the transformation matrix.Experiments show that the method can effectively correct the slanted X-ray images and make the BGA solder balls arrays in corrected image are aligned properly.Secondly,aiming at the problem that traditional methods are susceptible to interference from wires,chips and other objects around solder balls when dealing with BGA solder balls extraction under complex background,a method to segment BGA solder ball‘s region based on coordinate system transformation and edge detection is proposed.Considering the quasi-circular feature of solder balls,this method uses coordinate system transformation to map ROI region containing solder balls to polar coordinate system,then gets contour data by edge detection.At last,the original image is mapped back through the inverse transformation of the coordinate system.Experiments show that the proposed method of solder balls region segmentation based on coordinate system transformation and edge detection has the advantages in both accuracy and adaptability,and does not depend on parameter setting.Lastly,in the study of identification and extraction of voids in BGA solder balls,considering the low contrast of voids and the ambiguity of boundary,it is difficult to identify and extract voids accurately.A new algorithm for identification of voids in BGA solder balls based on Blob analysis is proposed.In this method,a series of binary images are obtained by using multiple thresholds to segment the balls region.Then,blob analysis based on the geometric characteristics of connected regions is used to identify the voids.At last,forest data structure is used to merge the recognition results to obtain the voids in solder balls.Compared with the single threshold method,the proposed method has better accuracy and adaptability.