NEW DATA ON THE EVALUATION OF THE INFRARED (IR) SPECTRA OF SUBSTANCES OF COMPLICATED STRUCTURE AND THEIR APPLICATION FOR IDENTIFICATION WITH PRIMA PATTERN RECOGNITION METHOD. PART I

In our paper the change in chemical composition was studied on samples (leaf, tendril- flower, stem, complete plant) taken in a 2-week period (July 2-July 16) of the development of tobacco plant, grown by two different modes of cultivation (bed and ensilage) with classical analytical and infrared (IR) spectroscopic methods. IR spectra were evaluated by traditional spectroscopic method and by pattern recognition method PRIMA (pattern Recognition by Independent Multicategory Analysis). The intensity of the IR bands was used for our investiga- tions, which was given in relative % value, introduced by us. Data needed for mathematical processing were obtained by a new procedure, by feature selection resting on spectroscopical basis. The dependence of cellulose and total nitrogen content, further the intensity of the IR spectral bands characteristic of these components, on the ripeness of the plant, on the quality of the plant part, on the mode of cultivation and on the place of the leaf on the plant were established. Various tobacco plant samples, having an identical IR spectrum on visual inspec- tion, were separated and identified with the PRIMA pattern recognition method with a recognition power between 97 and 100% according to their origin (ripeness, mode of cultiva- tion quality, place on the plant)