Classifying wetland areas

Ref-Nr:

Technology abstract

This model is presented for classifying wetland areas. In this project we address a new and very important issue: the observation of small backcountry wetland areas surrounded by different areas, hosting important species and delivering essential ecosystem services and biodiversity. Selecting the image sources multispectral channels in satellite imaging sensors for which backcountry wetlands can be best discriminated from the local neighbourhood;

Technology Description

Different type of images captured by various sensors were used for analyzing the temporal changes in landscape. The goal of the study was to get image information from multiple sources at different length scales in order to measure important features describing the state of lands. Aerial images were taken by a Sequoia multispectral camera mounted on a 3D Robotics Solo quadcopter. The Sequoia camera has an integrated GPS and a sunshine sensor, furthermore imaging is made by four narrow band and a RGB digital sensor. The images taken by the narrow band and RGB sensors were used to calculate the Normalized Difference Vegetation Index (NDVI) of the agricultural lands at different dates. To evaluate the given area from another perspective, images captured by the Sentinel satellite were also used for further analysis. The satellite images have 13 channels including the visible, near infrared and short wavelength infrared frequencies. The NDVI indices of the area were also calculated from the satellite images, and were compared to the results of the aerial measurements. 
A Multi-Layer Fusion Markov Random Field (ML-FMRF) model is presented for classifying wetland areas Wetlands play a major role in Europe's biodiversity. Despite their importance, wet lands are suffering from constant degradation and loss, therefore they require constant monitoring. This technology description presents an automatic method for the mapping and monitoring of wetlands based on the fused processing of laser scans and multispectral satellite imagery. Markov Random Field models have already shown the successful integration of various image properties in several remote sensing applications. In this description we propose the Multi-Layer Fusion Markov Random Field (ML-FMRF) for classifying wetland areas, built into an automatic classification process that combines multitemporal multispectral images with a wetland classification reference map derived from Airborne Laser Scanning (ALS) data acquired in an earlier year. Using an ALS-based wetland classification map that relied on a limited amount of ground truthing proved to improve discrimination of land cover classes with similar spectral characteristics. Based on the produced classifications, we also present an unsupervised method to track temporal changes of wetland areas by comparing the class labelings of different time layers. During the evaluations, the classification model is validated against manually interpreted independent aerial orthoimages. The results show that the proposed fusion model performs better than solely image based processing, producing a non-supervised/partly supervised wetland classification accuracy of 81-93% observed over different years.
Issue: the observation of small backcountry wetland areas surrounded by different areas, hosting important species and delivering essential ecosystem services and biodiversity. Although these patches are small one by one, but together they can contribute to the wetland cover area with a very high rate – their protection and mapping is a need. These small ecosystems act a main role in the Hungarian biodiversity – these many small patches may give an important contribution to the ecosystem in the Carpathian basin.
Selecting the image sources multispectral channels in satellite imaging sensors for which backcountry wetlands can be best discriminated from the local neighbourhood;
 Classifying the  sample areas of known natural reserves in order to create ground truth samples based on annotated satellite image data. Semi-automatic annotation will be performed by local clustering of living patches based on biological investigation;
 Defining the best Remote Sensing features which indicate backcountry wetland areas and contribute to their efficient monitoring process.
 

Innovations & Advantages

We will integrate classical RGB and multispectral cameras with aerial platforms (quadrotor and fixed wing) into an operational system for border protection applications. Moreover, we also utilize the freely available ESA Sentinel satellite imagery. The limited resolution of the satellite images is enhanced by capturing high resolution video footage of the target area. Namely, we co-register images from different sources and scales (resolutions). Secondly, our goal is to detect changes between images captured by fixed installed field cameras and images collected by aerial platforms in order to detect any unusual event close to the border.
The novelty is that we use new geometrical transformations to register UAVbased and terrestrial cameras in surveillance tracking and recognition, using different modalities (like RGF, Infra, Radar).
We will also use our new change detection algorithms, based on stochastic fusion methodology.
Selecting the image sources multispectral channels in satellite imaging sensors for which backcountry wetlands can be best discriminated from the local neighbourhood;
 - Classifying the  sample areas of known natural reserves in order to create ground truth samples based on annotated satellite image data. Semi-automatic annotation will be performed by local clustering of living patches based on biological investigation;
 - Defining the best Remote Sensing features which indicate backcountry wetland areas and contribute to their efficient monitoring process.
 

Current and Potential Domains of Application

The proposed combined spaceborn and airborne technology can be used for surveillance and safety, mainly for border security.