in the new approach by removing the need to
normal image and automatic determining the number
of partial images to be deleted, the speed of defect
detection increases and mean while maintains the
accuracy. In this paper quantity of partial images was
inspected in 3 modes: normalized, non – normalized
and quantized with the threshold value of 5. Three
variances were calculated for each mode and all
partial images. The best deletion algorithm was where
variances of partial images were calculated in non –
normalized mode. That is, normalizing and
quantization causes the variances to be impressed and
selection of proper partial images form them becomes
impossible. All together, the amount of SNS in this
approach has increased in comparison with other
approaches.