Since the 1950's character recognition has been an active field of research for computer scientists worldwide. The main reason is that character recognition is not only an interesting area of theoretical research with relevance to many pattern recognition sub-fields, but also a very needed and useful real life application. Making computers able to read would allow for substantial savings in terms of the costs for data entry, mail processing, form processing and many other similar situations. Every realistic character recognition system requires a feature extraction step in order to properly operate. This book is a large-scale review of the feature extraction approaches for character recognition based on literature review and experimental results. An original classification system is described, which groups feature extraction methods depending on their theoretical approach. The developed classification system aids in comparison and analysis of the feature extraction methods.