METHOD AND EQUIPMENT FOR AUTOMATIC HANDWRITTEN CHARACTER RECOGNITION USING STATIC AND DYNAMIC PARAMETER

摘要

PURPOSE: To provide a method and device for recognizing a handwritten character in response to an input signal from a handwriting converter. CONSTITUTION: Tablet data are generated and processed by the three independent sets of characteristic vectors for encoding shape information among input character information. Those characteristic vectors include the bit-mapped horizontal (x-axial) and vertical (y-axial) slices of input character data, and additional characteristic vectors for encoding absolute (y)-axial displacement from the base lines of the bit-mapped image. A recognition error due to a spatial or static processing is totally different from the recognition error due to a timely or dynamic processing. Moreover, those differences can be mutually compensated. As a result, the overall recognition error ratio can be sharply reduced by the connection of those two sources of the characteristic vector information. A method for connecting probability scores from the dynamic and static character models is provided.