Applied Engineering in Agriculture · 1996 · 15 citations · 0 references
Life SciencesBotanyPlant SciencesComputer VisionPlant AnatomyPlant MorphologyAgricultural Machine VisionCrop ClassificationLeaf MorphologyMorphological CharacterizationPlant IdentificationStem MorphologyBayesian AnalysisFuzzy Logic ClassificationImage-based Plant IdentificationLeaf Blade MorphologyMachine Learning For Plant IdentificationMachine Vision SystemsPixel-based Image AnalysisPlant Classification Systems
A method was developed for the automatic identification of the morphology of simple structured plants using machine vision. The method combined Bayesian pixel analysis, based on the {r,g,b} color space, and fuzzy logic, using both color and geometric parameters, to distinguish among primary plant parts such as main stem, petiole, and leaf blade. Three zonal geranium varieties were used to assist in method evaluation. Results indicated the combination of a constrained Bayesian classifier, allowing for uncertainty in portions of the {r,g,b} space, and a fuzzy logic classifier, based on partitioned decision rulesets, provided robustness in distinguishing between stem and leaf blade material across the three geranium varieties. The utilization of fuzzy logic enhanced the automatic decision generation process by giving a simple yet structured method of logic for the development cycle and decreased reliance on heuristics.