AUTOMATED MACHINE VISION GUIDED PLANT MONITORING SYSTEM FOR GREENHOUSE CROP DIAGNOSTICS
A machine vision guided plant sensing and monitoring system was designed and constructed to autonomously monitor and extract color features (Red-Green-Blue, Hue-Saturation-Luminance, and Color Brightness), textural features (Contrast, Energy, Entropy, and Homogeneity), morphological feature (Top Projected Canopy Area), plant indices (from NIR band relationships with color bands), and plant thermal radiation. A total of 17 features were extracted to monitor the growth and development of lettuce plants growing in a Nutrient Film Technique (NFT) system. From these 17 features, it was found that only 14 features were significant markers to separate the treatment group (induced water stress by withholding nutrient solution from selected rows) from the control. Using these 14 features at a 99% confidence interval and detecting when half of the features are shown to be significant as a threshold for onset of induced stress, it was shown that the system was able to detect the water stress 2 hours before human visual detection.
Story , D. and Kacira, M. (2014). AUTOMATED MACHINE VISION GUIDED PLANT MONITORING SYSTEM FOR GREENHOUSE CROP DIAGNOSTICS. Acta Hortic. 1037, 635-641
computer vision, crop monitoring, data acquisition, greenhouse, image processing