Enhancing Lidar Data using Generative Adversarial Networks
LIDAR (light detection and ranging) is an optical remote-sensing technique that measures the distance between sensor and object, and the reflected energy from the object. Over the years, LIDAR data has been used as the main source of Digital Elevation Models (DEMs). DEM have been used in variety of applications like road extraction, hydrological modeling, and surface analysis. LIDAR data accuracy and density are critical for reliable, accurate and high-resolution DEM generation. With recent development in Graphical Processing Units (GPU) and novel algorithms, deep learning techniques have become attractive to researchers for their performance in learning features from high resolution datasets. Numerous new methods have been proposed such as Generative Adversarial Networks (GANs) to create intelligent models that corrects and augments large-scale datasets. In this research, power of the GANs are explored to build intelligent systems that understand the issues within DEMs and improve the quality of the LIDAR datasets. Besides the correction, capabilities of GANs in improving the resolution of given DEM or removing noise and objects from DEMs to get bare earth products are studied.
