Thematic Accuracy Assessment for Object Based Classification in Agriculture Areas: Comparative Analysis of Selected Approaches

Over the past few years, the significant increase of request for spatial information is observed. Land cover and land use dat abases are needed for land management, urban planning, agriculture, environment protection and many other purposes. New sensors are able to provide very high spatial, spectral and time resolution data, what made the representation of earth surface more accurate but at the same time more difficult to classify in more objective and recurrent way. The object-based classification becomes frequently alternative approach to traditional pixel-based analysis. The paper presents the results of comparative analysis of different approaches applied to thematic accuracy assessment process of land cover classification with object-based methodology. The tests were done for the terrain with the diverse level of spatial and spectral complexity. The examples of images with di fferent resolutions, including VHRS one, were used in the study. The different variants of thematic accuracy assessment strategies were exploited for the results of object based classifications. Considering certain input assumptions, the results showed the pos itive aspects as well as limitations of applied approaches to thematic accuracy assessment.