Image Processing Methods for Detecting Voids in the Solder Joints of Surface-Mounted Components

Abstract

Void formation in the lead-free solder joints of electrical components can significantly affect their mechanical properties and thermal performance. Void-detection methods have been investigated in this paper, which are able to identify and measure voids in meniscus-shaped solder joints of surface mounted components automatically. In the experiment, 0603 size resistors (1.5 × 0.75 mm) were soldered onto a testboard, and X-ray images were acquired about the samples then. The image processing methods for the void detection were implemented in OpenCV-Python. At first, the properties of the images were analyzed (for example, calculating the histogram), and then the area, including the resistor and solder pads, was extracted by masking. Several methods were investigated for identifying voids in the images; Canny edge detection, global-and adaptive thresholding, and blob detection algorithm. The accuracy of the extraction methods was evaluated by identifying voids on a test image manually and comparing the results to that provided by the different methods. Canny edge detection was the best in our case, but global thresholding and blob detection are also promising solutions.