Deep learning as an alternative to super-resolution imaging in UAV systems
Anand Deshpande, Prashant P. Patavardhan, Vania V. Estrela, Navid Razmjooy, D. Jude Hemanth
Abstract
This chapter proposes a framework to super-resolve the low-resolution (LR) images captured using the unmanned aerial vehicle. The framework used a convolution neural network to super-resolve the LR image. This framework also removes the haze present in the LR image. The proposed system is evaluated using peak signal to noise ratio, structural similarity (SSIM) and visual information fidelity (VIFP) in the pixel domain. The experimental results demonstrate the advantage of the proposed method when compared to other state-of-the-art algorithms based on qualitative and quantitative analysis. Future trends in super-resolution (SR) unmanned aerial vehicle (UAV) imaging are discussed at the end of this chapter, followed by the concluding section.