Research on Flaw Detection of Transparent Materials Based on Polarization Images

Huafu Deng,Jianghua Cheng,Tong Liu,Bang Cheng,Kangcheng Zhao
DOI: https://doi.org/10.1109/iccasit50869.2020.9368546
2020-01-01
Abstract:Research on defect detection of transparent materials should first consider how to eliminate the effects of strong ambient light. The common polarization image processing method is to calculate the total light intensity, degree of polarization, et al. According to the difference in the polarization characteristics of the object surface, calculating the reference value of the polarization characteristic can improve the contrast between the object defect and the background. However, when dealing with the defects of transparent materials under complicated ambient light, this method will be greatly limited. This article proposes a new method to solve such problems. The algorithm can reduce the influence of ambient light and enhance defects on the surface and inside of transparent materials. The main content of the study is divided into two aspects: (1) Quickly obtain polarization images of transparent materials in different directions, complete the acquisition of polarization images, and then obtain the polarization characteristic parameters. (2) Considering the two directions of the spatial and frequency domains of the image, two methods of eliminating the ambient light of the polarization image are used to obtain a defective image with better image quality.
What problem does this paper attempt to address?