Aplicação do algoritmo haar cascade em um sistema embarcado para detecção de ovos do mosquito aedes aegypti em palhetas de ovitrampas

Abstract

Dengue is a world health problem, highlighting in tropical climate countries, especially in Brazil. There are still no effective forms of mosquito control, however, the use of traps called "Ovitrampas" can be done to detect the presence of it and to do the monitoring of the disease. In the Ovitrampas, the mosquito lays its eggs in a reed contained in a container with infusion of colonião grass, then this reed is subjected to microscopic and manual counting. However, this procedure is susceptible to failure and can culminate in errors. Thus, this work aims to automate the count in order to avoid possible failures, for this we use a set of hardware and software specific to the solution. The hardware has in its composition a raspberry PI and a coupled camera module whose goal is to capture a sequence of images from the vane, controlling the brightness in order to perform the processing of the image more effectively. The software is made up of imaging and detection algorithms to, using the images captured by the hardware, detect the presence of eggs in the reeds, we made was used from the OpenCV library to facilitate the implementation of the software. In preprocessing we use blurring and sharpening filters to improve the images. For detection we opted for the Haar Cascade algorithm, quite explored in the literature. Finally, we’ve dilated 10 reeds for the experiments. We compared the recognized eggs and accounted for by the proposed system with the actual amount obtained by the microscope in order to prove the effectiveness of the system.