【摘要】 提出一种基于小波分析与神经网络复合模型的图像特征提取方法.利用二维离散小波变换对图像信号进行滤波和边缘提取,作为图像的输入量供BP网络识别分析.构建了3层BP神经网络模型对图像特征进行识别,采用模糊加权算子的模糊化分层,特征提取模板提取图像中的不同特征,输出层采用均方差权值输出.通过对由50幅图像组成的训练集合进行训练和实验验证,结果表明,本方法的分辩误差率在4%以内.更多还原

【Abstract】 An identification method of image character based on the composite model of wavelet analysis and neural network is suggested.The two-dimensional discrete wavelet transformation is used to filter image signal and extract its brim character which is input into BP neural network for identification.3 layers of BP neural network are constructed to perform image character identification.The network is layered based on the fuzzy weighting operator which is composed by the fuzzy degree of membership.The feature matrix is formed by feature extraction pattern.The square-variance weight is used to improve the output.The designed neural networks firstly is trained by the training set of 50 images and then performs experimental verification for the testing set.The results show that this is a very effective distinguishing method and its resolution error is within 4%.更多还原