Source printer identification using printer specific pooling of letter descriptors

The digital revolution has replaced the use of printed documents with their counterparts. However, many applications require both due to several factors, including challenges security, installation costs, ease use, and lack expertise. Technological developments in domain have also resulted easy availability high-quality scanners, printers, image editing software at lower prices. Miscreants leverage such technology develop forged that may go undetected vast volumes documents. These mandate research on creating fast accurate systems for source printer identification We extensively analyze propose a printer-specific pooling improves performance local texture descriptor two datasets. proposed performs well using simple correlation-based prediction instead complex machine learning-based classifier achieving improved under cross-font scenarios. system achieves an average classification accuracy 93.5%, 94.3%, 60.3% Arial, Times New Roman, Comic Sans font types respectively, when only Cambria are available training.