Super-Resolution is the image processing technique to restore high-resolution images from low-resolution images.
In this work, a fast and low-latency super-resolution system has been implemented using a field-programmable gate array (FPGA), a device on which users can implement arbitrary logic circuits.
The system is based on a convolutional neural network (CNN). Unlike common CNN processors, the whole network is mapped on an FPGA as a pipeline, making the best use of parallelism for better performance.