Development of feature extraction techniques from large point clouds, hyperspectral imagery and their data fusion
Description of the research topic
“In the last years large datasets have been acquired by various remote sensing campaigns including airborne and terrestrial laser scanning (ALS and TLS, resp.), airborne and satellite hyperspectral (HS) remote sensing, photorobots scanning (PhRS) in man-made tunnels, etc. often financed by public resources, including governmental grants, EU research funding and alike. The projects were successfully completed, the data have been evaluated for a specific purpose, but, as a by-product, many other features, including geological, geomorphological, agricultural, biological, ecological features or built-up structures are also mapped, surveyed, scanned in the data, that are usually not evaluated. The current trend of data processing, the (semi)automated feature extraction allows processing of large data volumes; smart technologies pave the way for extraction, classification and categorization of features if ground truth data are available for a small portion of features.
The successful candidate (m/f) will develop effective computer codes and working computer applications processing large point clouds, HS data (a.k.a. “”””data cubes””””) and their georeferenced data fusion. The focus is on the processing of data fusion of data of various scales, e.g. ALS+HS+ground truth data of forests, scrub, wetlands, ALS or TLS or PhRSdata of geological and geomorphological features. Various methods will be tested (e.g. agent-based methods, random forest, etc.) in order to achieve an effective feature extraction and high-quality, practically applicable classification. The research question also includes
applicability and performance tests on various types of erroneous data, including misaligned georeference, effects of downscaling and differences in the measuring instruments of the fusioned data. The research is aimed at portability of the solution, whether the methods are applicable for data measured in different regions (flatland, hilly area), and under varying illuminaton conditions, and target-sensor angle configurations.”
Thesis supervisor: Balázs Székely
Required language skills: English
Further requirements: Good command of a programming language (e.g.Python), ability to use GIS and image processing software; experience in e.g., optimization, multi-dimensional data processing, feature extraction, classification is a plus.
How to Apply?
If you are interested apply here:[PhD] Doctoral School of Environmental Sciences – Eötvös Loránd University (elte.hu)
For more information visite the following website:Doctoral School of Environmental Sciences (elte.hu)
Funded: Not Funded
Master Degree: Required
Duration: 4 Years
Full/Part Time: Full Time
Starting Date: 06 September 2021
Deadline to Apply: 31 May 2021
Only local Hubs members can access this page. Join the community today: https://phdhub.eu/register
Fields of Science: