Solder joint's surface recovery based on linear hybrid shape-from-shading

Abstract

Solder joint's defect detection is a critical step in Printed Circuit Board (PCB) assembly. The inspection technology of solder joint's defect has developed from 2D to 3D. This paper deals with the recovery of the solder joint's surface based on shape-form-shading. Compared to other methods, this method has a very simple environment demand and has yet to be fully investigated. Generally, solder joint's images contain specular reflectance component and can not be recovered directly by conventional SFS methods, which relies on the assumption of Lambertian reflectance model. So hybrid reflectance model is introduced firstly. Secondly, we have deduced a discrete linear algorithm in hybrid reflectance model. In order to define the proportion of different reflectance components in hybrid reflectance model, we have developed a new computing method based on simulated annealing algorithm. Here, we describe the proportion as reflectance probability. Finally, Experimental results reveal that the proposed method shows better performance.