Automatische Erkennung von zehn Unkrautarten mit Hilfe digitaler Bildverarbeitung und Fouriertransformation

Abstract

Abstract Automatic identification of 10 weed species in digital images using Fourier descriptors and shape parameters Plant species discrimination in mixed plant communities has recently become possible using transforms and shape parameters to classify digital images. In the present study image analysis techniques were used to identify weeds commonly found in winter cereal fields. Those species included Veronica hedenfolia L., Thlaspi arvense Beauv., Alopecurus myosuroides L., Apera spica‐venti L., Poa annua L., Stellaria media L., Capsella bursa‐pastoris L., Lamium purpureum L., Matricaria chamomilla L. and Galium aparine L. Images of several growth stages of these weeds were photographed using a Still Videokamera, binarified, the shape extracted and then Fourier descriptors and shape parameter were calculated for each weed. Classified digital images of each species were stored on the computer. A separate set of photographic images of these 10 weeds were used to test the ability of the classified images for plant identification.The average rate of correct identification was 81.9 % ranging from 41.6 % to 100 %.