Object Oriented Classification – a comparative study of two ENVI Feature Extraction methods

Being considered a process that requires fewer resources, extracting features of interest from satellite imagery may prove to be an alternative that can provide good results while have a low production cost and high applicability. This study aimed to analyze two Objects Oriented Classification methods implemented on ENVI software to suggest which method is much feasible in classifying a satellite image of a complex urban area. To achieve the best result in the classification process were used ancillary data (nDSM, indices and masks). Based on the results: accuracy, visual inspection, time spent for each classification process, resources cost, etc., the ENVI classification methods showed they're power in thematic maps production