Optical Character Recognition for Automated Cartography: The Advanced Development Handprinted Symbol Recognition System

摘要

This NORDA Technical Note is composed of five chapters. The first chapter presents an overview of optical character recognition (OCR) and its relation to the automated cartography environment. It provides the DMA Subtask objectives and discusses them in the light of symbol digitizing and information transformations. The division of a total OCR system into data acquisition/document management and isolated character recognition is considered along with NORDA's recent tasking (FY-82) prototype for DMA production centers. Chapter Two presents a discussion of the different ways in which recognition systems are constructed. In particular, it considers the differences in approach necessary for constrained and free-form OCR. Chapter Three describes the DMA environment in which a handprinted OCR system must operate and discusses performance requirements. The general structure of the Handprinted Symbol Recognition System is outlined in Chapter Four. This material considers the key issues of information content, problems in the thinning or vectorization of a character, shape measurement and feature extraction, and finally character recognition or labeling. The interaction between each of these elements is emphasized. Chapter Five provides a brief summary of the current Subtask accomplishments and status along with areas where work is in progress toward developing other handprinted OCR capabilities for DMA.