Wavelets analysis for defects detection in flat glass

In this paper we present a multiresolution approach for the detection of defects in flat glass. This approach is based on a signal restoration using 1D wavelet transforms. In a first step a procedure is applied to automatically determine the best reconstruction level based on the energy distribution of wavelet coefficients. By properly selecting the smooth subsignals or detail subsignals for reconstruction, the repetitive patterns can be effectively removed and only local anomalies are preserved in the restored signals. Experimental results show that defects in flat glass are well detected. A superposition of all treated lines allows to reconstitute the images revealing clearly the defects