This Interdisciplinary, Integrated, Interactive (I3) capstone project, Automating change detection in ImageQuerying, built upon a command-line Sentinel-2 download script developed for the Application Development IP using Python and the downloading agent aria2. Driving factors were the desire to improve Python skills by further automating a complete workflow, and a developed interest in semantic enrichment of EO data from the Remote Sensing Analysis and Modelling seminar, exploring automatic knowledge-based spectral categorization with the Satellite Image Automatic Mapper™ (SIAM™) by Dr. Andrea Baraldi. The workflow is visualized in the flowchart (Fig. 1), as are some of the changes detected over a year of a dam being built in Asyut, Egypt (Fig. 3) based on the semantic query (Fig. 2) of 33 SIAM™ generated semi-concepts.

AIQ Workflow

Figure 1: Workflow displaying what was automated, semi-automated, and manual

Figure 2: semantic query constructed in IQ identifying pixels changing from vegetation or built up to water (blue), and vice versa (red)

Figure 3: map of semantic spatio-temporal query results from IQ

All code for the project can be found on GitHub (AIQ @augustinh22). The project is continuing to be developed, with work underway to enable the download of all ESA Sentinel satellite data by developing a platform-independent, open-source QGIS plugin.

The full text of the project report can be found here. Below is a copy of the introduction and selected references:

Earth observation (EO) data are being rapidly collected at increasing spatial and temporal resolutions, faster than researchers can utilize and with free and open data volumes outpacing existing spatial data infrastructures. The age of free and open EO big data has arrived and methods of incorporating semantic enrichment as well as more efficiently storing, accessing and querying them, whether spatially, temporally or semantically, are being explored. Increased inclusion of semantics transform ssuch big data into meaningful information. Many spatial analysis routines can be chained together and automated, yet analysis of remotely sensed images often requires more user interaction to produce meaningful results.

This research project is situated in an overall vision to develop an automated workflow for post-classification change detection analysis based on Sentinel-2 EO images, the Satellite Image Automatic Mapper (SIAM™) developed by Dr. Andrea Baraldi and the cutting-edge ImageQuerying (IQ) system primarily using Python. IQ is an automatic near-real time EO image understanding system (IUS) with the goal of enabling multi-temporal, spatial and semantic queries based on image content rather than exclusively based on metadata (Sudmanns et al. 2016; Tiede et al. 2016a; Tiede et al. 2016b). These queries could eventually include situations like monitoring effects of forest fires over time in a relatively large, remote area, looking at seasonal changes in vegetation or water bodies over a multiple-year time frame, or identifying areas with low to no cloud cover at specific times regardless of overall cloud cover of the original images loaded into the database.

Tiede et al. (2014) were able to automate post-classification change detection using Landsat images. The concept of this project is to do similar change detection analysis, but using Sentinel-2 images and pre-classified categories, also known as semi-concepts, rather than land cover classes. Incorporation into the IQ array database facilitates a range of possible queries and makes automation considerably more dynamic instead of automating a specific query. Two test case scenarios were chosen to illustrate the workflow. Automating routine EO data acquisition and analysis tasks can be seen as an efficient approach to tackling EO big data challenges. The results of this project contribute to existing efforts under development at Z_GIS, University of Salzburg’s Interfaculty Department of Geoinformatics.

Selected References

Sudmanns, M., Tiede, D., Augsten, N., Baraldi, A., Belgiu, M., Lang, S., 2016. Array-Datenbanken für semantische inhaltsbasierte Suche und Analyse in Satellitenbildarchiven, in: Dreiländertagung Der DGPF, Der OVG Und Der SGPF in Bern, Schweiz – Publikationen Der DGPF, vol. 25. pp. 555–564.

Tiede, D., Lüthje, F., and Baraldi, A. 2014. Automatic post-classification land cover change detection in Landsat images: Analysis of changes in agricultural areas during the Syrian crisis, in: Seyfert, E., Gülch, E., Heipke, C., Schiewe, J., Sester, M. (Eds.), Band 23: Geoinformationen Öffnen Das Tor Zur Welt, 34. Jahrestagung in Hamburg 2014. Publikationen der Deutschen Gesellschaft für Photogrammetrie, Fernerkundung und Geoinformation (DGPF) e.V., Potsdam.

Tiede, D., Baraldi, A., Sudmanns, M., Belgiu, M., Lang, S., 2016a. ImageQuerying – Earth Observation Image Content Extraction & Querying across Time and Space, in: P. Soille and P.G. Marchetti (Eds.) Proc. of the 2016 Conference on Big Data from Space (BiDS’16), EUR 27775 EN, doi:10.2788/854791, 2016. pp. 192–195.

Tiede, D., Sudmanns, M., Baraldi, A., Belgiu, M., Lang, S., 2016b. ImageQuerying – prototypische Implementierung eines Systems zur raumzeitlichen und inhaltsbasierten Satellitenbildsuche und Satellitenbildanalyse. AGIT ‒ Journal für Angewandte Geoinformatik, pp. 80–85.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s