AUTOMATED FEATURE EXTRACTION ALGORITHMS FROM MULTISPECTRAL SPACE BORNE DATASETS

Many current and future applications in spatial science will require proper management of resources demanded for spatial and temporal information about terrestrial targeted objects acquired from multiple sources. According to the literature, pace of urban development and consequent sprawls demands an analysis of the current spatial and temporal data that can lead to a better understanding of infrastructural needs, land use planning, imperviousness and monitoring of water and air pollution. Tremendous effort has to go into helping to monitor environmental changes in cities that has increasing population growth, which results into the expanding of urban sprawls such as newly develop housing facilities, more denser congestion of traffic, flooding, more commercial business areas that all leads to reduction in air and water quality. Urban environments are also dynamic and undergo rapid physical and socioeconomic changes. These environmental changes can only be accurately study and monitor by using satellite imagery data, which give real time results and provide a view over the area of interest.

Managing resources demands spatial and temporal information about terrestrial target objects. The paper describes innovative methods of extracting target objects from worldview-2 MSS Dataset over the Bronx Borough in New York. Spatial and spectral attributes were computed to build algorithms that were used to extract different sets of target objects from the imagery. The algorithms were tested on a study area and the accuracy of the classification was estimated using the confusion matrix model. Results show that several target objects from urban areas such as the Bronx may be extracted in near-real time from satellite data. The approach described in the study may be applied to any geographic region that has similar characteristics to the Bronx.