An LSI hardware design for online character recognition using associative memory

In this paper, we propose an associative memory based system for real-time character recognition. Based on an associative memory with 128 reference patterns of size 256 bits designed in 0.35 μm technology we could get an average nearest-match search time of 130 ns for classification of different samples of characters written in Times and Arial fonts. Comparing to other OCR systems, although this prototype model is not yet robust enough, it is advantageous in terms of classification lime and hardware size.