All-in-focus synthetic aperture imaging using generative adversarial network-based semantic inpainting

Highlights：

• The first time to address the missing information problem caused by heavy occlusion in Synthetic Aperture Imaging.

• Our method generates a realistic all-in-focus synthetic aperture image where the information of the occluded region is completely restored.

• Extensive experiments on public datasets and our own datasets demonstrate the superior performance of the proposed method over state-of-the-art Synthetic Aperture Imaging methods.