Counterfeiting of currency globally remains a significant problem. And according to the authorities, a large portion of the fake currency is produced by Small Office Home Office inkjet printers. Therefore, a new inkjet printer forensics technology would be useful to identify the model of the source printer given a print sample. In our paper, we study the print patterns from 15 low-cost inkjet printers that are being sold on the market and examine test targets at the microscopic level. We design 4 printer intrinsic features including Dot Size, Dot Density, Average Distance to Nearest Dot, and Nearest Dot-Sector Density Function to characterize the behavior of inkjet printers. Furthermore, we extend our research and develop a machine learning based Printer Identification System. Unlike handcrafted features that have intuitive meaning to human viewers, an alternative set of intrinsic features are extracted from the Residual Neural Network, and based on the Neural Network features, a Support Vector Machine classifier is trained and is able to perform the printer model classification. Our evaluation shows that the proposed system produces robust and reliable results.