Fragment Polygon Removal in Incremental Land Cover Map Updating

Land cover and its changes can reflect the natural and human activities on the Earth's surface, as well as the social development and natural change in a specific period of time. The classification of remote sensing images is an effective approach for generating land cover maps. Two main strategies for producing continuous multiphase land cover maps are available: classifying each phase image separately and developing a new phase map by using two-phase images to extract land cover changes and update products, that is, incremental updating. In incremental updating, only the target area that has changed is updated. Incremental updating has the characteristics of minimal processing time and easy data transformation. The accuracy of the incremental updating of land cover products is mainly affected by two aspects: 1) the classification accuracy of land cover change patches and 2) the splicing fit between the change patches and the baseline map to be updated. In this paper, the fragment polygons caused by the second issue are discussed, and a method is proposed to obtain a new phase of land cover classification map via patch mosaic. The land cover data of Jackson City, Mississippi, USA is used as an example. The effect of fragment polygon on incremental updating is explored, and quantitative indicators are obtained.