An evaluation of an object-oriented paradigm for land use/land cover classification

Object-oriented image classification has tremendous potential to improve classification accuracies of land use and land cover (LULC), yet its benefits have only been minimally tested in peer-reviewed studies. We aim to quantify the benefits of an object-oriented method over a traditional pixel-based method for the mixed urban-suburban-agricultural landscape surrounding Gettysburg, Pennsylvania. To do so, we compared a traditional pixel-based classification using maximum likelihood to the object-oriented image classification paradigm embedded in eCognition Professional 4.0 software. This object-oriented paradigm has at least four components not typically used in pixel-based classification: (1) the segmentation procedure, (2) nearest neighbor classifier, (3) the integration of expert knowledge, and (4) feature space optimization. We evaluated each of these components individually to determine the source of any improvement in classification accuracy. We found that the combination of segmentation into image objects, the nearest neighbor classifier, and integration of expert knowledge yields substantially improved classification accuracy for the scene compared to a traditional pixel-based method. However, with the exception of feature space optimization, little or no improvement in classification accuracy is achieved by each of these strategies individually.