摘要
在人类手写教码识别中,脱机不规则数码的识别是识别算法中比较难解决的问题.本文提出的一种识别方法,将笔道密度识别算法和层次特征识别算法相结合,对不规则数码脱机识别起到了十分有效的作用.在对0-9共10个数码,用多种方法手写,经扫描、二值化、平滑、边缘跟踪、压缩、细化等处理后,它的识别正确率在90％以上,达到了很好的效果.
In recognizing number written by hand, the recognition of offline irregular number is more difficult to solve. The paper proposes a kind of recognition method, which combines track density recognition algorithm with level feature recognition algorithm, and it is very effective for offline irregular number. After scanning, two value adjusting, smoothing, edge trailing, compressing and thinning , its identification correct rate is more than 90% for 0 - 9 numbers written with various methods.
出处
《小型微型计算机系统》
CSCD
北大核心
2003年第5期940-942,共3页
Journal of Chinese Computer Systems
