Object-Based or Polygon Accuracy Assessment
This chapter presents a number of approaches and the corresponding considerations to conduct an accuracy assessment when using a polygon-based classification approach for generating a thematic map. As in any thematic accuracy assessment, evaluating an object-based or polygon map requires sample units of reference data to compare with the map so that an error matrix can be generated. The chapter deals with the many factors that must be considered when conducting an object-based image analysis (OBIA)-based accuracy assessment, including what is the appropriate sample unit for the reference data and what methods should be used for the collection of the samples. When the OBIA approach is used to generate a thematic map, pixels with similar spectral characteristics are grouped into segments/objects, and then, the segments are classified as a group instead of pixel by pixel. By far the biggest consideration when conducting an object-based accuracy assessment is the determination of the sample unit.