| Height is a good indicator of plant growing status and also a predictor of yield potential. An automatic plant height sensing system can save time and labor costs. The objectives of this study were to evaluate the performance of three range sensing technologies including ultrasonic sensor, laser range finder, and range camera on plant height measurement, and to recommend an approach which has potential of realizing infield by-plant height measurement. Data acquisition systems were developed to control sensors and collect data both during laboratory and field tests. Sensing objects were corns and sorghums at early growth stages. Data and image processing algorithms were developed for height information extraction, by-plant recognition, and error calculation. The errors between manually measured ground truths and sensor measurements using ultrasonic sensor, laser range finder, and range camera were 16.2%, 12.4% and 18.9%, respectively. The ultrasonic sensor featured with low cost, robust in-field performance; however, it was too sensitivity to the orientation of sensing surface. The laser range finder gave the highest measurement accuracy and was robust and stable in field application. Though it can hardly give the by-plant information by itself because of its line scan nature, it had the potential of realizing by-plant height measurement by being used together with other techniques such as side-view sensing. The range camera had the lowest accuracy which might largely due to the errors induced in the image registration process. It was also very sensitive to the outdoor sunlight and needed proper shading. However, if a good by-plant top-view recognition algorithm could be developed, the range camera has the potential of realizing by-plant height measurement if only one sensor would be used to sense from the top view. |
