Philippine E-Journals

Validation of Lepidopteran Classification by Morphological Fractal Dimension Analysis

Anna Rose P. Nillama | Mark S. Borres | Ricky B. Villeta | Roramie V. Arco | Joyce C. Unabia

Discipline: Bio-Diversity

 

Abstract:

The paper uses fractal geometry through the fractal dimension of photos of the Lepidopteran species, the commonly named butterflies, collected from the regions of Central Visayas, Philippines to determine if the morphological dimensions can discriminate one species over the other for classification purposes. The images of species were converted into a binary image using the simplest form of image processing known as thresholding effect. The binary images provide a simple view and ease for computing its fractal dimension. The fractal dimension explains the space – filling property of the image. The species available for use were the Cethosia, Idea, Pachliopta, Catopsilia, Troides where 10 subspecies of each were chosen. Also, the male species were isolated and tallied separately from the females. Empirical results revealed that the fractal dimensions can differentiate the species from one another with a general p – value of 0.000 and F – value of 6.483. All sample pairwise comparisons were tested for the significant differences and still differentiates one over the other as evidenced by the corresponding p - values of all equal to 0.000, respectively. Moreover, the fractal dimensions, in relation to the species gender, were tested for possible significant differences, and findings showed that it could differentiate across butterfly’s gender with a p – value of 0.000 and F- value of 8.037. The empirical probability of wrong classification using the fractal dimension is less than 1%.