Abstract: The void extraction method of solder joints based on global threshold segmentation is inaccurate. Moreover, once the solder joint in the image is shaded, the extraction of BGA solder joint's voids will become difficult. In order to detect the void defects of BGA solder joints under complex X-ray imaging background, it was detected that the unobstructed solder joints based on thresholding, solder roundness and solder area, and an interactive method of X-ray contour extraction to detect the obstructed solder joints was propose. According to the characteristics of solder buddle, an integrated method of solder joint's void extraction was presented based on grayscale morphology, histogram stretching, fuzzy enhancement and BLOB analysis. The result of comparison experiment shows that the method presented in this research possesses high accuracy in detecting the void defect of BGA solder joints.