Análise do cultivo de alface crespa em ambiente protegido baseada em grandezas físicas e processamento digital de imagens

Abstract

Lettuce is the most consumed vegetable in the world and has handled R$ 8 billion just in the retail markets over the year of 2013 in Brasil. A good part of that source is thanks to the modern agriculture, application of technologies, that had helped to maximize production, productivity and to reduce costs. Digital image processing, alongside sensors and actuators, could help to reach the best moment for harvesting, allowing a better profitability for producer and avoiding possible losses, such as waste of water, fertilizer and time. Today, the process is highly dependent on human perception, and premature or late harvests may occur. The objective of this work was to develop monitoring systems of physical quantities, digital image processing and acquisition. For the hardware development: sensors DHT11, LM35DZ, DS18B20 and LDR were used alongside a WebCam (Logitech C920); actuators: solenoid valve and lamps; and microcontrollers: Arduino and Raspberry PI. The OpenCV library was used for the digital imaging processing, and, through the Java programming language, for pre-processing (color space convertion, image correction through dilatation and erosion mathematic morphology) and the processing (image segmentation and classification of variety), and, from the obtained amount of pixels, variety development analysis was carried out, with the help of gaussian and linear function adjustments. Two experiments were carried out, one from june to august and another one from october to december 2018, both totaled 66 days of cultivation (21 days of sowing and 45 days of development in a greenhouse) at the experimental area of the Universidade Federal da Fronteira Sul (UFFS) in Laranjeiras do Sul,PR campus. While, from the agronomic point of view, the winter variety Palmas differs from the others on total fresh mass, the summer varieties Palmas, Crocante and Pira Verde significantly differ on leaves fresh mass from Simpson and Grand Rapids varieties. The results obtained through the function adjustments had shown the effective possibilities for digital image processing on helping at lettuce development proccess, correlating the amount of pixels with the growth lettuce foliage over time.