LORENCS, A., MEDNIEKS, I., SINICA-SINAVSKIS, J. 2014. Simplified Classification of Multispectral Image Fragments. Electronics and Electrical Engineering. Kaunas: Technologija, 20(6), pp. 136–139.
A simplified approach to classification of multispectral image fragments by their specific spectral features is presented. Application of this approach to discrimination of vegetation areas occupied by the Giant Hogweed species is described and compared with an approach based on calculation of the Consolidated Covariance Image. The proposed method is based on calculation of mean and standard deviation and successive thresholding within certain spectral bands that are found to be informative for the specific task by analyzing the ground truth data. It is shown that the method provides close to perfect discrimination of Giant Hogweed from other vegetation areas represented in ground truth data (absence of commission errors together with clear identification of Giant Hogweed fragments in corresponding ground truth regions). Simplicity of the method provides for fast processing of multispectral images from large areas. The proposed approach is perspective for analysis of multispectral images in different application fields where it is possible to choose several informative spectral bands, e.g. in biomedical imaging.