Determination of the most suitable color space for intelligent water stress discrimination for plants inside the greenhouse (Case Study: Coleus)

A precise estimation of required water for plant depends on many factors, among which the percentage of ground cover is a key parameter. Digital image processing and machine vision can be widely used to obtain this parameter in irrigation management applications. The aim of this study was to recognize the required water for plants based on color parameters of plants’ ground cover; therefore, different color spaces (RGB, rgb, XYZ, HSV, HLS, L*a*b, L*u*v*, YCbCr, YUV, TSL and I1I2I3) were applied on the set of ornamental shrub images with the scientific name of Plectranthusscutellarioides in two positions (fresh and wilting).Each color space demonstrated different probability distribution ofa given color class corresponded to two plant conditions (fresh and wilting). Thus, after examining the color spaces, both statistically and visually, the suitable color spaces were selected. Finally, histograms of suitable color spaces have been used to train the Bayesian Classifier. The Bayesian classifier detected two conditions of plant (fresh and wilting) with precision 83.11%. In general, on the basis of information obtained from images histograms, (frequency of pixels’ intensity) plantswater status for irrigationschedulingwas recognizable.