Facsimile signal processing for Chinese characters

A facsimile (FAX) signal processing system designed for filtering documents with printed Chinese characters is described. Filtering algorithms are applied to remove noise corrupting the image scanned by the FAX machine and to preserve the edges. A total of six filtering algorithms have been chosen for investigation, namely, weighted average filter, separable median filter, median with checking filter, Lee's filter, gradient inverse filter, and rotating variance filter. Documents containing Chinese characters are scanned and then stored in the RAM of a personal computer. Random noise is then added, and filtering performed. Experimental results show that all of the six algorithms are capable of removing some of the noise. However, the median filter demonstrated the best noise-removing and edge-preserving capabilities, speed of operation, and ease of implementation for both small- and large-size Chinese characters.