A highlight processing technology for SMT solder joint gray image

The gray image of the SMT solder joint usually contain specular highlight component, which is a serious obstacle for reconstructing the three-dimensional shape from SMT solder joint based on shape from shading (SFS) theory. Thus, a highlight processing method for the gray image of SMT solder joint is presented. This method consists of two parts. One is highlights detection to gray images. It is based on the Bayesian decision theory, which is regarded as a special kind of image segmentation with the Markov random field as a priori conditions, and the gray-scale data as likelihood conditions, and then split the highlight region from non-highlight one with the iterative conditions model (ICM) algorithm. The other is image restoration. to make fully use of the information of the solder joint shape, a complementary color method for the highlight removal is put forwarded, which uses mean field surface normal continuity theory and interpolate the highlight areas with the knowledge of local mean surface normal of non-specular areas. In the end, two experiments can be done to compare the gray images of original one and processed one; the 3d reconstruction of images also can be done. the results show that our approach can effective handle highlight component in the gray images of SMT solder joint, meanwhile, the method is efficient and stability. The further meaning is that it has an important significance for the three-dimensional reconstruction of the SMT solder joints.