Main Article Content
It is shown that by the method of remote registration of fluorescent radiation excited by laser radiation, it is possible to determine not only the species of plants, but also to divide classes of plants belonging to subspecies within the class. To achieve accuracy, discriminant analysis was applied and the quality of the classifier was evaluated.
Full-scale flight experiments performed by us using onboard laser measuring equipment to map the level of fluorescent signals showed distinct differences in their spectral forms depending on the state of ground vegetation, although in a series of these experiments the signal excess over the noise level was not too high. These measurements demonstrated the possibility of confident remote diagnostics of stress conditions of terrestrial vegetation based on the registration of fluorescent responses.
The clarity of such a description consists in the fact that the data characterizing a particular class form a point (or a collection of points) in a multidimensional feature space, with dimension n - equal to the number of selected channels. Then the task of interpreting remote measurements is reduced to determining whether the n -dimensional vector of the spectral response belongs to a particular class. The degree of reliability and quality of identification is ultimately determined by how clearly these classes will be separated in the space of spectral features.