Fractal Dimension-based Drosophila melanogaster Wing Genotype Classification

摘要

Experts may use biometric techniques in identifying biological images. For example, the genus Drosophila has been used as a genetic model organism for more than 100 years. Facial recognition and geometric morphometry, which requires human interaction, are methods used in the processing of images. In this paper, we propose the following method of combing stationary wavelet transformation, Canny filter and fractal dimensions to classify the genotype of fly wings of the Drosophila melanogaster species. We validated the methodology by testing various classifiers and varying the test and training samples. We achieve a higher hit rate than the comparative work by using a reduced number of attributes and an automatic algorithm. The use of only female samples significantly increased the accuracy.