Fuzzy decision system for threshold selection to cluster cauliflower plant blobs from field visual images

Current work deals with the design of a fuzzy reasoning system capable of selecting the most appropriate threshold to binarize a sequence of cauliflower crop images. Binary images speed up the processing time to detect the crops and calculate their position. The proposed threshold algorithm is based on a performance criterion, the emergence of a binary image wherein cauliflowers are easily distinguished from the background and no relevant parts of the plants are lost. Once the best threshold is applied and the binary image is obtained, a fuzzy clustering algorithm has been designed to rejoin the fragmented blobs of the plants and recover the cauliflower image. Finally, plant positions are calculated based upon the area and location of the blobs belonging to a plant. This technique is extremely valuable for real time precision agriculture tasks, as are those of selective spraying.