The criteria of choosing the optimal Gabor filter and defect detection using the optimal Gabor filter

This paper proposes the criteria of choosing the optimal Gabor filter and uses the optimal Gabor filter to detect defections for glass fiber clothes. On one hand, by using eight direction templates which replace the filters' eight directions the authors zehong chen and xiaoxia feng have choosed two optimal orientations. However, we just choose one optimal orientation by computing the maximum of the sum of the whole pixels. On the other hand, we choose the optimal scale through the image's information obtained from the frequency domain and the disturbance coefficient which is based on the theory of eigenvalue perturbation. The optimal scale is in front of the scale that disturbance coefficient is minimum. In other word, that the scale making disturbance coefficient minimum subtracts one is the optimal scale. Finally, experimental results demonstrate that our proposed approach is feasible and effective.