结构引导的图像修复

针对粗网络引入先验知识较少使得补全的内容存在明显视觉伪影问题，提出了基于边缘结构生成器的两段式图像修复方法。采用边缘结构生成器对输入的图像边缘和色彩平滑信息进行特征学习，生成缺失区域的结构内容，以引导精细网络重构高质量的语义图像。通过在公开的图像修复基准数据集Paris Street-View上进行实验测试，结果表明，所提模型可对掩膜占比达50%的图像进行补全。在客观的量化评价指标上，峰值信噪比、结构相似度系数、L₁和L₂均值误差等数值整体优于EC、GC、SF等方法，其中，掩膜占比为0%~20%时，峰值信噪比指数达到33.40 dB，优于其他方法2.37~6.57 dB，结构相似度系数提高了0.006~0.138。同时，补全的图像纹理更清晰，视觉质量更高。 

Images inpainting via structure guidance

Aiming at the problem of obvious visual artifacts in the content of rough network with less prior knowledge, a two-stage image inpainting method based on an edge structure generator is proposed. The edge structure generator is used to perform feature learning on the input image edge and color smoothing information, and generate the missing structural contents so as to guide the fine network to reconstruct high-quality semantic images. The mentioned method has been tested on the public benchmark datasets such as Paris Street-View. The experimental results show that the proposed approach can complete the hole images with the mask rate of 50%. The quantitative evaluation indicators: PSNR, SSIM, L₁ and L₂ errors respectively surpass current images inpainting algorithms with excellent performance, such as EC, GC, SF, etc. Among them, when the mask rate is 0%-20%, the PSNR index reaches 33.40 dB, which is an increase of 2.37-6.57 dB compared to other methods; the SSIM index is increased by 0.006-0.138. Meanwhile, the completed images get clearer texture and higher visual quality.