Combined object-based classification and manual interpretation–synergies for a quantitative assessment of parcels and biotopes

Abstract

Recent technical advances in remote sensing data capture and spatial resolution lead to a widening gap between increasing data availability on the one hand and insufficient methodology for semi-automated image data processing and interpretation on the other hand. At the interface of GIS and remote sensing, object-based image analysis methodologies are one possible approach to close this gap. With this, methods from either side are integrated to use both the capabilities of information extraction from image data and the power to perform spatial analysis on derived polygon data. However, dealing with image objects from various sources and in different scales implies combining data with inconsistent boundaries. A landscape interpretation support tool (LIST) is introduced which seeks to investigate and quantify spatial relationships among image objects stemming from different sources by using the concept of spatial coincidence. Moreover, considering different categories of object fate, LIST enables a change categorization for each polygon of a time series of classifications. The application of LIST is illustrated by two case-studies, using Landsat TM and ETM as well as CIR aerial photographs: the first showing how the tool is used to perform object quantification and change analysis; the latter demonstrating how superior aggregation capabilities of the human brain can be combined with the fine spatial segmentation and classification. Possible fields of application are identified and limitations of the approach are discussed.

Keywords:

• image segmentation
• object characteristics
• object fate
• spatial analysis
• visual interpretation
• LIST extension

Acknowledgements

The work conducted has been financed by the EU 5th framework projects SPIN (Spatial Indicators for European Nature Conservation, Contract No. EVG2-2000-0512) and Iron Curtain (Contract No. QLRT-CT-2001-01401). We kindly acknowledge the fruitful discussion with Prof. Ulrich Kias and Walter Demel from the FH Weihenstephan. We gratefully thank the Berchtesgaden National Park and the company Geospace for providing the data sets.