Water stress is the major environmental stresses that affect agricultural production worldwide, especially in arid and semi-arid regions. Substantial increases in yield could be possible if irrigation water was applied at the most appropriate time to prevent excessive drought stress. Therefore, information about plant water status for irrigation scheduling is necessary. Two important physiological indices for detecting water stress are stomatal conductance (g s ) and relative water content (RWC). However, the determination of these indices is time-consuming and labor-intensive. In this research, the feasibility of processing RGB digital images of sesame ( Sesamum indicum L.) canopies for g s and RWC determinations were examined. A greenhouse experiment was conducted on potted plants of two genotypes of sensitive (Naz Takshakhe) and resistance (Yekta) of sesame; subjected to three water stresses: well-watered (WW), medium-water stressed (MWS) and severe-water stressed (SWS) using a factorial experiment within a completely randomized design. The canopy images under control light conditions were obtained using a 24-bit RGB digital camera (16.1 megapixel, HX 100 Sony), placed in a zenithal position. A threshold-based approach was applied in plant detection based on image segmentation by Keywords: Water stress, Image processing, Artifical neural network, Relatie water content, ltr"
