F. Adamo, F. Attivissimo, N. Giaquinto, M. Savino
A NEW TECHNIQUE FOR GLASS QUALITY CONTROL
The paper deals with the problem of pointing out and then measuring the manufacturing defects (scratches and halos) of handmade satin glasses by an accurate, real-time and cheap method. A wavelet-based de-noising technique is used to obtain surface information about glasses under inspection. The proposed algorithm first obtains a two-dimensional wavelet transform of a high resolution image of the glass under inspection which, according to multi-resolution analysis, produces a suitable number of decomposition levels of the image, and then it carries out a thresholding operation on details. Finally, using the threshold levels estimated considering the present noise level, it assesses and characterizes the defects. The final aim of the work is the realization of an automatic and in-line computer vision system for glasses analysis; it must be able both to accurate inspection and to quality degree assessment of the final products as they come out from the production chain.