Abstrakt
A system for recognition of woody species in Central Europe according to the images of their leaves is described. Our own data set, which includes 151 species at this moment, with at least 50 leaves per species was used.
After segmentation, the contour of the leaf was traced. Fourier descriptors normalised to translation, rotation, scaling and starting point of the boundary, were used.
The size of the leaf, if known, was used as a separate feature. The nearest neighbour classifier was used.
The algorithm is available through a web application.