Abstract
Inspired by the surprising performances of deep generative models, in this paper we present the preliminary results of an overly ambitious task: estimating computationally the additional spectral bands of a color aerial image. We have harnessed the expressive power of deep generative models to estimate the distribution of mostly infrared bands of aerial scenes, using only color RGB channels as input. Our approach has been tested from multiple aspects, including the reconstruction error of the additional bands and the effect of estimated bands on scene classification performance, as well as through the transfer potential of the trained network to a distinct dataset. To our surprise, the initial experiments have shown us that deep generative models can indeed learn to estimate additional bands up to a certain degree and can thus computationally reinforce datasets stemming from color-only sensors.
Highlights
Thanks to advances in sensor technology, the spectral resolutions of remote sensing images have increased to unprecedented levels, paving the way for new applications and improving the performances of existing ones
There are various definitions and mathematical models for generative models, in essence they assume that some observed variable x that is mapped to a label Y by a discriminative method is produced by a hidden process or variable z with some unknown distribution p(x, z)
Original images of the UC Merced [14] (UCM) dataset are directly downloaded from United States Geological Survey (USGS) databases and resized into 256 × 256 pixel images
Summary
Thanks to advances in sensor technology, the spectral resolutions of remote sensing images have increased to unprecedented levels, paving the way for new applications and improving the performances of existing ones. High spectral resolution is known to improve the performance of a wide range of critical remote sensing applications, ranging from target detection [1] and scene classification [2] all the way to pixel classification [3,4,5,6,7], through the availability of complementary information Even though such multi- and hyperspectral image acquisition devices are nowadays more accessible than even before, they are still not as widespread as RGB color sensors. There are various definitions and mathematical models for generative models, in essence they assume that some observed variable x that is mapped to a label Y by a discriminative method is produced by a hidden process or variable z with some unknown distribution p(x, z)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: TURKISH JOURNAL OF ELECTRICAL ENGINEERING & COMPUTER SCIENCES
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
