This document discusses techniques for extracting person names from scanned documents with diverse genres and noisy optical character recognition (OCR) text. Six techniques are compared: using dictionaries of names, regular expressions, context-free grammars, maximum entropy Markov models, conditional random fields, and an ensemble that combines dictionary, regular expression, and Markov model extractors. The dictionary and regular expression techniques achieved the highest performance according to evaluations on a test set of documents annotated with full names. The authors conclude more advanced techniques could build on this initial work to further improve name extraction.