Research Of Solder Joint Detection Based On Computer Vision

With the development of information technology, the level of automation production has become higher. In the electronics industry, with the development of the refinement and automation, solder joint’s quality profoundly affects the electronic product’s quality and efficiency. The defect inspection of solder joints has become one of the focuses of the electronics manufacturing industry. The traditional manual detection method, not only time consuming, but also low accuracy and low efficiency, and kinds of new methods gradually came into being. The application of machine vision to automation production and solder joint detection has become the trend of modern automated production. With the development of the manufacturing industry, China has shift from manufacturer to produce power, it puts forward the high requirements for intelligent test for solder joint’s automate test. And automatically, flexible and intelligent manufacturing has become the inevitable developing trend of welding technology.Machine vision, which applied in the solder joint detection, through the solder joint image collection, fixing the solder joint area, and extract the solder, analysis the image characteristics, and the parameters will be to feedback to the end actuator, can greatly reduce the artificial error, improve the production efficiency and ensure the quality and reliability. In the process of production, use of machine vision module and digital image processing algorithm to test the quality of solder joint, can be used in many electronic components. Such as PCB solder joints, mobile phone GBA solder joint, etc.In this paper, we research the solder joints of cell phone lithium battery. Combine machine vision and image processing method to realize the defection of leak welding, solder joint adhesion, virtual welding.In this paper, the main work includes:First of all, this paper introduces the commonly used image preprocessing and feature extraction methods, according to the characteristics of the solder joint image, an improved median filter is introduced to the gray-scale image to enhance the contrast and remove noise in images. Put forward a reasonable fast iterative method of binary image segment.Secondly, on the basis of full consideration of gray scale, space relationship and algorithm is complexity, on the basis of this paper proposes a watershed segmentation algorithm based on distance transform, which can accurately segment the target area and adhesion area.Then, characteristics of solder are detected and identified. This article focus on four types of defects:empty, welding bridge and insufficient solder joints, and leak welding.Finally, the system control software is developed to realize the effective detection of solder joint, and the experimental results are presented.