Research of glass defects detection based on DFT and optimal threshold method

Abstract

In this paper, we proposed a defect detection algorithm based on Discrete Fourier Transform (DFT) and optimal threshold method, mainly aiming at detecting bubble, stone, and crack of glass. Firstly, the test images are pre-processed using median filter. Then the defect region is highlighted by spectral residual approach in Discrete Fourier Transform (DFT). Finally, the optimal threshold, using which the defect region is segmented, is determined by multiple iterations. Experimental results demonstrate the proposed detection algorithm can detect and localize the defect region with high accuracy and better practicability.