Abstract

A new application of fractal dimension analysis was initiated for identification of different kinds of broadleaf weeds. The distribution of chlorophyll time series of weed leaves exhibited self-similar geometricalcharacteristics. The fractal dimension analysis was conducted using Grassberger-Procaccia (G-P) phase space reconstruction algorithm. A total of 300 leaves of three weeds of Oxalis corniculata L. (OC), Ixeris chinensis (Thunb.) Nakai. (IC) and Herba glechomae L. (HG) (species) were sampled and analyzed. Thecorrelation dimensions of time series of O. corniculata, I. chinensis and H. glechomae estimated by G-P algorithm are 8.050, 10.094 and 11.730, respectively. The distribution of chlorophyll time series was restricted in chaos environment and governed by strange attractors. The self-similar distribution property includes the information about weed varieties, which can be used for classification.

Key words: Fractal dimension, identification, chlorophylls, weed, Grassberger-procaccia (G-P) algorithm.