Method and apparatus for character preprocessing.

摘要

An optical character recognition system comprises a raster scanner (1), a video detector (2) and an encoder (3) for coding video data into a compressed run-length code. Compressed video data is stored by means of a direct memory access unit (DMA, 10) into a memory area (22) in a memory (13). A control computer (14) is building up a start-of-scan table (23) pointing at compressed video data in said memory area (22). Further, the computer is boxing a one character area by setting up a box pointer (B) in the start-of-scan table and a horizontal displacement in a register area (24). When a character area is boxed the computer proceeds to expand the boxed data and transfers it into a character recognition area (25). Correction means (29) are used in the control computer to readjust the box as required.

The run-length code is a code for storing only changes of black to white and white to black data instead of storing all video bits. The expanded code is a code for storing all ' video bits.