A Frequency Domain Constraint for Synthetic and Real X-ray Image Super Resolution

Synthetic X-ray images are simulated projected from CT data. High-quality synthetic can facilitate various applications such as surgical image guidance systems and VR training simulations. However, it is difficult to produce high-quality arbitrary view in real-time due different slice thickness, high computational cost, the complexity of algorithms. Our goal generate high-resolution by upsampling low-resolution with deep learning-based super-resolution methods. Reference-based Super Resolution (RefSR) has been well studied recent years shown higher performance than traditional Single Image Super-Resolution (SISR). It fine details utilizing reference but still inevitably generates some artifacts noise. In this paper, we introduce frequency domain loss a constraint further improve quality RefSR results without obvious artifacts. To best our knowledge, first paper for functions field super-resolution. We achieved good evaluating method on both real datasets.