Research On Solder Joints Inspection Based On Image Surface Recovery

Solder joints quality inspection is an important means of ensuring the quality of electronic products. Generally, solder joints quality inspection includes the component defect detection and the detection of solder paste printing quality. Concretely, solder joints quality inspection usually refers to the solid-state solder joints for testing after reflow. There are many solder joints quality inspection methods, currently the most popular means of detection is based on machine vision for automated optical inspection (AOI).Solder joints quality inspection should meet the following key problems: the problem of image acquisition, feature extraction of solder joints, discrimination issues about solder joints quality. Although a large number of solder joints quality inspection equipments have been designed in market, but they generally only work in 2D testing mode, can not make a more objective assessment for solder joints at the level of three-dimensional quality information. This paper studies these issues, proposed a solder joints quality inspection method based on the recovery of the surface of solder joints. Contents include four parts:For the image acquisition in the solder joints inspection system, in order to meet the requirement of random location during the user operating process and the information recording of large components, the panoramic image acquisition of printed circuit board is finished based on the scale independent features transformation; in order to improve the efficiency of image acquisition of solder joints, a method based on path optimization is proposed. This method makes use of the profile data obtained from the printed circuit board designing stage to plan the path offline, which changes the traditional online programming model. Therefore, it greatly simplifies the programming effort before detection.For the feature extraction of solder joints, the surface recovery technique is introduced into the solder joint quality inspection firstly. However, there are often more serious specular highlights in solder joints image, which will directly affect the solder joint's surface recovery results. So, a surface reflectance component estimation method based on the surface shape distribution is proposed, under the dichromatic reflection separation model, diffuse and specular reflection components are estimated effectively. On this basis, two different solutions are proposed: one is to propose a illumination-constrained inpainting method based on the assumption of curvature continuity, remove the specular reflection component in the image to reduce or remove the effect of specular reflection on the surface recovery; another way is to introduce the Blinn-Phong reflection model, which is a hybrid reflection model, and under this model to improve the surface recovery algorithm. A linear hybrid shape-from-shading algorithm is proposed based on Blinn-Phong reflection model. Through the effective estimation of reflection components, not only the reflection component will be specified automatically, but also makes recovery accuracy is improved.For the discrimination issues about solder joints quality, the characteristics of the solder joints are analysised firstly. For some features that are easy to quantified, such as component missing, excessive offset, solder bridged, solder angle of climb and the volume of solder, the criteria will be defined. Then through the measurement based on the solder joints image, these key feature's data will be obtained and applied to evaluate ts quality. Finally, the concept of tin volume index is proposed to make the detection about tin volume defect quantified. This rule-based method of determining the quality of solder joints has the advantage of clear physical meaning, a quantified determined result can be given. However, false defects for tin, tin volume index is only from the viewpoint of tin volume, not from the surface shape of solder joints for qulity analysis. So, this paper proposed a detection method based on the shape index, using support vector machine to classigy the different tin colume solder joints. Experimental results show that the using different quality discrimination methods for different characteristics of solder joints can be a more reasonable assessment.In summary, this paper proposed a solder joint quality detection method based on the surface recovery techniques, resolved the problem of solder joints image acquisition and solder joints surface recovery under the conditions of specular reflection, as the same time, three-dimensional features extraction of solder joints and quality discrimination has been identified.