Integrating Multispectral Satellite Images with High-Resolution Imagery for Object Classification

One of the urgent tasks arising in the development of geographic information systems is the development and application of new methods that allow the creation and updating of geographic databases based on space images and other sources of information to be considered. Object-oriented classification methods based on image segmentation are gaining interest in obtaining thematic maps suitable for direct storage in Geographic Information System (GIS) databases. The limitations of effective application of image segmentation are often related to the spatial resolution of the images used. The presented paper proposes a solution to this problem based on using images of different spatial resolutions. World View-1 (WV1) panchromatic image with high spatial resolution (0.50 m) and Landsat Thematic Mapper (TM) multispectral image with medium spatial resolution (30 m) were used to perform object-based land cover classification in the study region. The proposed methodology is a testament to our meticulous approach, involving a sequential application of segmentation and classification techniques. The TM satellite image was classified using the Maximum Likelihood classifier in the initial stage. Subsequently, the World View-1 image was segmented using a region-based segmentation algorithm. Finally, the segmented WV1 image was classified, with the previously classified images from the TM satellite image serving as a benchmark. To ensure the utmost accuracy, the final land cover map was verified through a comprehensive field survey.