To enhance the effectiveness of font family generation,this paper proposes a method that integrates SCConv and attention mechanisms,specifically targeting the unique characteristics of font families by adjusting the network to focus on local font features and stroke intersections. This method replaces standard convolution with spatial-channel reconstruction convolution (SCConv) to improve the efficiency of network generation. Additionally,the model incorporates shuffle attention (SA),which combines and groups font features to increase feature interaction,and uses a multi-head attention mechanism to fuse the font features. Experimental results show that the proposed model outperforms the DG-Font algorithm in both metrics and visual outcomes for generating font family characters.
font familyfont generationattention mechanismsSCConv
万方数据
维普
