The objective of this study was to identify vessels in radar images obtained by the R-99 airplane from Brazilian Air Force. Amplitude data obtained in L-band and HH-, HV-, VH- and VV-polarizations over the region of Tubarão Harbor, State of Espirito Santo, were processed through different types of enhancements, filters, classifiers, and spectral transformers. The images with highest performances were also analyzed to differentiate between military and merchant vessels, considering the five elements of interpretation (shape, size, shadow, tone and associated factors, that is, the context in which the vessels are found in the images) and five steps of image interpretation (detection, reconnaissance, analysis, deduction, and classification). The best combination of image processing was 50-200 contrast enhancement, aperture or erosion filters, and SVM (Spectral Vector Machine) classifier or SCI (Synthetic Color Image) transformation. It was possible to discriminate vessels in the steps of detection and reconnaissance, while the differentiation between military and merchant vessels was possible in the steps of analysis and deduction. In the final step of classification, it was not possible to define the types of military vessels (e.g., frigate or destroyer) neither the types of merchant vessels (e.g., petroleum tanker or bulk carrier).
Remote Sensing; Characterization of Vesselso SAR Image Processing; Spectral Enhancement
