Microscopic printing analysis and application for classification of source printer

Identifying a forged printed document with scanned evidence can be challenge. Microscopic printing is showing random shape which depends on the source as well material. This paper presents statistical analysis of patterns under microscopic scale, analyses effect direction, substrate (uncoated and coated paper), technology (conventional offset, waterless electrophotography). The shows negligible yet, using descriptor indexes, materials technologies are distinguishable scale. As result, algorithms based Support Vectors Machine Random Forest developed, indexes features, for identification. Both proposed algorithms, equally, achieve high classification accuracy rate, over 92% complex geometric-shape patterns. Thanks to lightweight efficiency Machine, study promising applications real-world potential implementation in Internet Things devices.