摘要
针对汉字的多样性和相似性不同于西方字母,识别相对困难的问题,提出了基于ART神经网络的汉字识别方法.在识别前,利用OpenCV(开源计算机视觉库)将汉字进行图像处理,为后续识别提供输入数据;然后经ART神经网络对输入数据进行训练识别.采用8组相似度较高的汉字作为样本进行实验,证明了方法的有效性.
Outing to the differences between Chinese and western words and the difficulty in recognition, a Chinese character recognition model based on the ART neural network is proposed. Before recognition, the Chinese characters are processed by OpenCV (open source computer vision library), which provides input data for the following recognition procedure. Then, the input data are trained and recognized by ART neural network. In the experiment, eight groups of similar Chinese characters were used as samples for the sake of making the results universal. The results show that the recognition of the shape- changed Chinese characters among the provided samples is reliable and accurate.
出处
《天津科技大学学报》
CAS
2013年第4期74-78,共5页
Journal of Tianjin University of Science & Technology
基金
国家自然科学基金资助项目(50975204)
关键词
ART神经网络
汉字识别
OPENCV
不变矩
ART neural network
Chinese character recognition
OpenCV
moment invariance