Automatic plant identification with chlorophyll fluorescence fingerprinting

摘要

The development of precision farming needs methods for automatic identification of individual plant species. We have earlier shown that chlorophyll fluorescence induction curves can be reliably used for automatical identification of plants (Tyystjarvi et al., 1999). In the present study we show that a high accuracy of recognition can be obtained even if the teaching set for pattern recognition is collected several weeks before identifying a test batch of plants. It is also shown that very simple fluorescence traces can be used for the identification, and that dark pre-incubation of the plants can be shortened to a few seconds without seriously compromising the power of the method. The method is even more powerful if the aim is only to distinguish one crop species from weeds. The data shown here suggest that the fluorescence fingerprint can be developed to a method of practical importance for precision farming.