Projection profile and Fourier transform for Chinese character recognition

Chinese character recognition is normally time consuming and requires expensive computer systems. However, with the advent of microcomputer technology, and the development of processing techniques, on-line recognition of single font printed characters is now feasible. This paper studies the problem of on-line recognition of printed Chinese characters. An efficient procedure based on projection profile and Fourier transform using black point density (BI'D) classification is developed and tested. The test results indicate a satisfactory recognition rate while the algorithm requires modest storage space and processing time. The proposed technique is shown to be superior to other more sophisticated techniques generally used in character recognition.