摘要
文章将智能控制与模式识别方法相结合,提出了一种仿人识字行为的图像汉字智能识别机;采用决策控制级、模式识别级和反馈校正并行式结构。决策控制级选择最佳的识别策略和参数,模式识别级进行多分类器结果融合识别,并根据对模式识别级输出的可信度判断结果进行反馈,校正识别方法和参数,最终提高汉字的识别效率。
An intelligent recognizing machine of Chinese character images is presented in this paper. It combines intelligent control with pattern recognition. A parallel structure consisting of the decisionmaking level, the pattern recognition level and the feedback level is adopted. At the decision-making level, the best recognizing strategies and parameters are chosen. At the pattern recognition level, multi-classifier results fusion is carried out. At the feedback level, different results of recognition are fed back so as to revise the recognizing methods and parameters. This structure can increase the recognition efficiency.
出处
《合肥工业大学学报（自然科学版）》
CAS
CSCD
北大核心
2006年第10期1268-1272,共5页
Journal of Hefei University of Technology：Natural Science
关键词
反馈
识别策略
多分类器结果融合
feedback
recognizing strategy
multi-classifier results fusion
