基于改进生成对抗网络的书法字生成算法

李云红,段姣姣,苏雪平,张蕾涛,于惠康,刘杏瑞

Calligraphy generation algorithm based on improved generative adversarial network

Yun-hong LI,Jiao-jiao DUAN,Xue-ping SU,Lei-tao ZHANG,Hui-kang YU,Xing-rui LIU

表 1 内容编码器的网络参数

Tab.1 Network parameters of content encoder

网络层类型 网络参数 卷积层 4×4 Conv, 64 filters, 2 strides 卷积单元层 4×4 Conv-IN-LRelu, 128 filters, 2 strides 卷积单元层 4×4 Conv-IN-LRelu, 256 filters, 2 strides 卷积单元层 4×4 Conv-IN-LRelu, 512 filters, 2 strides 卷积单元层 4×4 Conv-IN-LRelu, 512 filters, 2 strides 卷积单元层 4×4 Conv-IN-LRelu, 512 filters, 2 strides 残差单元层 Resnet Block 1×1 Conv, 512 filters,1stride 卷积单元层 4×4 Conv-IN-LRelu, 512 filters, 2 strides 残差单元层 Resnet Block 1×1 Conv, 512 filters, 1 stride 卷积层 4×4 Conv-LRelu, 512 filters, 2 strides