Print Scum Identification using DCT based Computer Vision Method

This paper presents a computer vision approach for print scum identification using DCT (Discrete Cosine Transform). Scumming is a common print problem, occurs when non-image area accepts ink. In order to get better quality print, scumming is unwanted. Moreover in scumming, density difference of printed image and unwanted background pixels are not remarkably different, so it becomes difficult to identify the scum area from printed area by image intensity variation. In this paper this is achieved using frequency component reduction, to identify the area of scumming in several print samples, both text as well as image. In both the cases this technique gives considerable separation of scum pixels from the original printed image. The pictorial presentations of the results show the remarkable potential of the presented technique which can be a possible alternative of present subjective measure of scumming.