In this paper, we present a new system to segment and label document images into text, halftone images, and background using feature extraction and unsupervised clustering. Each pixel is assigned a feature pattern. The invariant feature pattern is then assigned to a specific region using the Expectation-Maximization (EM) algorithm. Once the segmentation is performed, a specific enhancement filter can be applied to each document component.
Mohamed N. Ahmed, "Image Segmentation Using Expectation Maximization and Its Application to Digital Copying" in Proc. IS&T Int'l Conf. on Digital Printing Technologies (NIP21), 2005, pp 412 - 416, https://doi.org/10.2352/ISSN.2169-4451.2005.21.1.art00021_2