Estimando a orientação das linhas de cana-de-açúcar em imagens aéreas

Abstract

With the growth of the world population, the demand for agricultural areas has reached the limits of sustainability. It is then urgent to increase the efficiency of plantations, producing a more significant amount of crop without increasing the planted areas. In this context, the area of Precision Farming has emerged, integrating state-of-the-art technologies into agricultural systems. One of the most demanded cultivars is sugarcane, used both as food and biofuel. In sugarcane harvesting, the precision of a few centimetres is required to prevent the cultivars from being crushed by the harvesters. The autonomous harvesters are guided by satellite positioning systems, whose precision can reach several meters. One proposed solution is to use unmanned aerial vehicles equipped with cameras and other sensors capable of providing real-time information to the harvesters. This work shows the development of image analysis for estimating crop orientation by two heuristic approach using some well-know processing techniques. To validate the solution architecture, an aerial image set was acquired and labeled by experts. To create a robust ground truth, a methodology for fusion of experts voting was used. With the ground truth, the estimating technique was validated, and its implementation was analyzed, presenting performance and quality within the application requirements.