Estudio comparativo de cámaras de bajo coste para su implementación en un nodo de visión para la monitorización de cultivos hortícolas

Laburpena

With the development of precision agriculture, new techniques are needed for the optimal determination of irrigation needs and crop status, being one of these techniques the use of vision nodes that allow the capture of images for later processing. Many previous studies have shown the benefits of using images in agriculture, although capture is usually performed in manual way or with high cost cameras, which makes it unfeasible to do a permanent installation. The objective of this article is to carry out a comparative study of the use of low cost cameras, instead of the standard cameras, for vision nodes in horticultural crops that automate the entire process, considering different environmental conditions. The evolution of a crop growth is analyzed using the images obtained with different low cost cameras. These cameras are connected to an Arduino board either by serial or parallel bus, that manages its sending by GPRS to a web server, where the images are processed for the determination of the percentage of vegetation cover by segmentation with several color models. Then, the main advantages and disadvantages of using the different cameras studied are indicated, taking into account the ambient luminosity in various resolutions to determine the optimal configuration. The position of the camera is also considered, for those situations where the placement of the camera in a zenith position is not possible, analyzing the capture of images depending on the angle and distance, the possibility of using the zoom or a manual focus that improves the quality of the captured images. With the data obtained, the requirements of different buses and communication protocols are extracted, detailing their processing time, memory and communication bandwidth requirements, as well as its possible implementation in wireless systems that allow the use of drones.