Morphological and textural plant feature detection using machine vision for intelligent plant health, growth and quality monitoring

A machine vision guided real time plant health/growth and quality monitoring system was designed and developed for use in controlled environment plant production systems. The developed system consisted of two main components including a robotic positioning system to place a color CCD camera on designated plants or on the canopy, and an image processing module. Extracted plant features to determine overall plant growth and health status included: (as morphological features) top projected canopy area (TPCA), plant perimeter, weighted radius, (as color features) Red-Green-Blue values, Hue-Saturation- Luminosity values, color brightness, (as textural features) entropy, energy, contrast and homogeneity. The capability of the system and feasibility of using these plant features for timely detection of lettuce tip burn were tested with an initial experiment and the results are presented. The monitoring system was capable of extracting plant morphological, textural and temporal features evaluated. The extracted plant parameters: top projected canopy area, entropy, energy, contrast and homogeneity showed some hopeful signals for detection of lettuce tip burn occurrence. However, experiments are in progress to further evaluate these plant features and system capability on early tip burn detection.