Evan W Krieger, Sidike Paheding
Download Project (1.5 MB)
Image interpolation has been widely used for enhancing spatial resolution of the input images. Generally, the spatial resolution enhancement techniques are categorized into single frame and multiple frame super resolution. Multi-frame super resolution techniques use a set of low resolution frames, while single image super resolution only requires one single input to reconstruct a high resolution image. In real life applications, single image super resolution is preferred when lacking of multiple frames in the data. In this work, we present a single image interpolation approach for reproducing high frequency missing components of the input low resolution images. The high frequency feature is first extracted in Fourier domain, and then the system is trained to regenerate better pixel values, which contribute to better resolution. We evaluate the method visually and quantitatively using several test images.
Vijayan K. Asari
Primary Advisor's Department
Electrical and Computer Engineering
Stander Symposium poster
"Image Interpolation Using Fourier Phase Features" (2016). Stander Symposium Projects. 822.