A new portable application for automatic segmentation of plants in agriculture

The achievement of the objectives of precision agriculture requires not only the development of new technologies, but also the availability of portable tools that can be used in the field. This paper describes the creation of a novel application for mobile devices called pCAPS (portable classification application for plants and soil) that integrates several computer vision techniques for plant segmentation and analysis in crop pictures. This tool allows monitoring of agricultural crops in real time, providing information that can be used to automate and optimize the calculation of water needs. The three main modules of pCAPS are: capture and cropping, image analysis, and historical record. First, a robust algorithm to detect rectangular markers located in the field is proposed; images are trimmed accordingly, in order to achieve a uniform analysis over long periods of time. Then color segmentation is applied using a probabilistic approach based on histograms in the optimum color space. Finally, an object counting process is performed on the binarized images, which is useful in applications that require the number of objects and their average size. Using pCAPS, the user can go with a portable device to a crop field, take a picture with the camera, automatically cut out the image, and perform on the ground an analysis of the vegetation, obtaining the percentage of green cover (PGC), number of plants, date, time, and GPS coordinates. This information can be sent by email to the central offices of the agricultural business, where appropriate decisions on fertirrigation needs would be taken. pCAPS is intuitive, user-friendly and has been developed for use by farm managers, requiring minimal skills in the use of mobile devices.