alexh@rtech.rtech.com (Alex Hwang) (07/24/89)
I am looking for information on COMMERCIAL printed Chinese character OCR product. Preferred hardware platform is IBM PC compatible. Feedbacks so far have been mostly negative. It is apparently difficult to develop. I am curious about the technical reasons for this. (I am almost a zero in character recognition topic. I would appreciate any references of articles on this topic.) The following is a short summary on the characteristics of Chinese characters: 1. A page of Chinese characters are printed in a 2 dimensional array of equal size rectangles. 1 character occupies 1 rectangle. (No proportional spacing concept.) 2. There are more than 10,000 distinct Chinese characters. 3. The average number of strokes of Chinese characters is more than 10. 4. Assume 1 font only. 5. There are no equivalent of Chinese spelling checker. Therefore, accuracy is very important. Chinese proofing and editing is very painful for non-experts. Does the fact that the Chinese characters have more strokes then English alphabets improve the accuracy potential in Chinese character recognition? Again, any info or comment is greatly appreciated. Alex Hwang