Development of a low-cost machine vision system for selective sprayer

In this paper, a novel system design methodology for developing a low-cost real-time machine vision system for selective sprayer was reported. The system consisted of multiple low-cost Universal Serial Bus (USB)-based video-conferencing digital cameras, a 16-bit microprocessor-based controller and a laptop PC. Sensor resolution under the consideration of minimum sensor detectable weed size was quantified through experiment. Based on sensor characteristics, corresponding system integration strategies and algorithms were developed. With incorporating low-cost sensors, the cost of sensing components will be dramatically reduced compared with traditional analog video camera and frame grabber-based machine vision systems. A prototype system was set up and the results of performance evaluation experiments validated the system design. Low-cost, simplicity and high-efficiency featured the developed system.