Defect detection in periodic patterns using a multi-band-pass filter

This study presents a rapid and reliable technique for the inspection of defects in a two-dimensional periodic image using a multi-band-pass filter. More importantly, the blurring effect of the resultant defect images is significantly reduced, thereby resulting in a more precise estimation of the size of defects when compared with methods that use low-pass filtering. As a filter-based approach, the present technique does not require an alignment procedure. In addition, computational time is reduced by implementing multi-band-pass filters with convolution masks when the filters are operated in the spatial domain. Further, this approach involves mostly addition operations with very few multiplications; hence, computational time is significantly reduced when compared with those for existing approaches. The efficiency and effectiveness of the proposed multi-band-pass filter is verified through examples. It is observed that there is a significant reduction in blurring effects, leakage effects, and computational effort. It is noteworthy that though the proposed approach is presented as a two-dimensional filtering problem, it can be reduced to a one-dimensional filtering problem under the assumption that the misorientation angle of the inspected periodic pattern is negligibly small.