geostatistics; NVDI; remote sensing; Landsat; colorization; algorithms
Rasera Luiz Gustavo, Gravey Mathieu, Lane Stuart N., Mariethoz Gregoire (2020), Downscaling Images with Trends Using Multiple-Point Statistics Simulation: An Application to Digital Elevation Models, in Mathematical Geosciences
, 52(2), 145-187.
Benoit Lionel, Gourdon Aurelie, Vallat Raphaël, Irarrazaval Inigo, Gravey Mathieu, Lehmann Benjamin, Prasicek Günther, Gräff Dominik, Herman Frederic, Mariethoz Gregoire (2019), A high-resolution image time series of the Gorner Glacier – Swiss Alps – derived from repeated unmanned aerial vehicle surveys, in Earth System Science Data
, 11(2), 579-588.
Gravey Mathieu, Rasera Luiz Gustavo, Mariethoz Gregoire (2019), Analogue-based colorization of remote sensing images using textural information, in ISPRS Journal of Photogrammetry and Remote Sensing
, 147, 242-254.
Nussbaumer Raphaël, Mariethoz Grégoire, Gravey Mathieu, Gloaguen Erwan, Holliger Klaus (2018), Accelerating Sequential Gaussian Simulation with a constant path, in Computers & Geosciences
, 112, 121-132.
A high-frequency and high-resolution image time series of the Gornergletscher - Swiss Alps - derived from repeated UAV surveys
This dataset is based on aerial photographs of the Gornergletscher glacial system (Switzerland) collected during ten intensive UAV surveys carried out approximately every two weeks throughout the summer 2017.The final products consist in a series of 10 cm resolution ortho-images, Digital Elevation Models of the glacier surface, and Matching Maps that can be used to quantify ice surface displacements.
G2S: The GeoStatistical Server
This project aims at using automatic colorization techniques for adding consistent spectral bands to archive satellite images. Recent colorization algorithms have proven very efficient for the automatic restoration of color in historical grayscale images. The foundation of this project is that these methods are equally applicable to adding several spectral bands to historical satellite images. To date, colorization algorithms have been strictly limited to the colorization of grayscale images. Moreover, these methods have remained in the realm of digital photography, with no attempts to apply them for the enhancement of scientific observations. This project will for the first time apply colorization to multispectral remote sensing images. As such, it will also be the first application of colorization for the enhancement of scientific data.One of the most important uses of remote sensing data is the assessment of changes in Earth surface features through repeated imagery. Using past remote sensing data is an ideal way to explore the variability induced by climate change or by direct human activity. Change detection approaches have been increasingly used in the last decade, with the advent of online portals providing free historical satellite imagery. However, despite the quantum leap provided by the existence of such “big data”, a major limitation of change assessment using satellite images is the relative shortness of the period where data are available. In this sense, we are historically myopic in our grasp of environmental changes. Satellite records of Earth surface features really started in 1972 with the launch of the Landsat-1 satellite, leaving us with a historical perspective of less than 45 years. In the context of a rapidly changing Earth, this is a very short observation period. Moreover, this historical depth is further reduced by the relatively low information content of earlier observations. Technological improvements in the sensors result in the documented record of the Earth surface being non-uniform with age. For example, early Landsat 1-3 satellites covering the period 1972 to 1983 were limited to 4 spectral bands, while more recent Landsat satellites provide information on 11 bands. This project will investigate the use of state-of-the-art colorizing algorithms to reconstruct the spectral bands that are missing on older sensors. This will involve the development of an automatized colorization method aimed at enhancing the spectral coverage of historical remote sensing images using textural information, and its application to the recovery of the missing near- and far- infrared bands in the 1972-1983 Landsat images. In addition, it will explore applications of the colorization of spectral bands (spectral enhancement), with in particular urban heat island change detection, the computation of NDVI by addition of spectral bands to images taken with low-cost digital cameras, and the use of added spectral bands to improve the results of unsupervised classification procedures.