Identification of the Aedes aegypti / albopictus Mosquito Using Digital Image Processing Techniques

Abstract

According to the World Health Organization, dengue is estimated to affects 390 million people a year. In Brazil alone in 2019, 1,544,987 probable dengue cases were reported. The main vectors involved in the spread of dengue are the mosquitoes of the species Aedes aegypti and Aedes albopictus. In addition to dengue they also transmit other diseases such as chikungunya, zika fever and yellow fever. One way to help the Zoonosis Centers to better control these dangers would be through a tool that could help in the identification of mosquito species. In this work we propose a methodology to classify mosquitoes by species and sex through the use of images associated with Digital Image Processing techniques. After acquiring mosquito samples, we created three databases, one using a smartphone camera, a macro camera and a Digital Single Lens Reflex camera. The Auto Color Correlogram, Gray Level Co-occurrence Matrix and Hu Moments techniques were applied to extract characteristics from the images, and then we used the Support Vector Machine, Random Forest and K-Nearest Neighbors classifiers. We classified 800 images divided into four classes and three databases. In our experiments, we obtained a maximum accuracy of 99.53%.