论文部分内容阅读
利用数据融合技术，对手写数字字符的识别进行数据级、特征级及决策级的融合．数据级融合首先对数字字符进行去噪、锐化处理，然后采用加权平均值算法来提高字符样本的信噪比；特征级融合采用特征点扇形提取法来提取字符特征，然后用BP神经网络法对字符特征进行学习和识别；决策级融合采用模糊逻辑法对最终的结果进行判定，形成了一种精度较高的数字识别方法．实验结果表明，数据融合方法相比传统的字符识别方法在识别效果上有较大提高．“,”By means of data fusion technology, this paper fused the data of handwritten numeral characters recognition on data level, feature level and decision - making level. The data fusion, began with noise elimination and sharpening to the numeric characters, and then, improved the signal to noise ratio through the weighted average algorithm, and extracted the features of the characters by feature level fusion employing feature point fan -shaped extraction, and learned and recognized the features of the characters by the BP neural network, and determined the final results with decision - making level fusion which uses fuzzy logical method. As a result, the data fusion formed a numeral recognition method with high precision. The experiments showed that the data fusion method, compared with the traditional numeric character recognition method, improved a lot in recognition effect.