The field of practical character recognition has seen great advances these years, owing the to proposition of effective newmethod and progress in computer hardware. However, lots of problems like precision and progressing speed remain to be solved. This paper proposes a useful practical optical character recognition method of both Chinese character and English character. Given a photo randomly taken, we pre-progress it to undo the influence of light and tilt. Then we separate each word to extract 8-direction chain-code feature of it. When faced with complex scene, the method we propose achieve relatively high precision and fast speed.
|Publication status||Published - Mar 2015|
CitationLu, P., Yang, Y., Sheng, B., Li, P., Chen, M., & Wu, D. (2015, March). Chinese character recognition based on 8-direction feature extraction. Paper presented at the 4th International Conference on Multimedia Technology (ICMT 2015), Sydney, Australia.
- Chain-code feature
- Tilt correction