5. Conclusion
The performance of an advanced real-time plant discrimination system, employing two red laser diodes (635 nm, 685 nm) and one near-infrared laser diode (785 nm) in conjunction with a linear sensor array, has been investigated experimentally. Experimental results have shown that the use of dual Normalised Difference Vegetation Indices (NDVI) in conjunction with data aggregation and threshold optimisation significantly enhances the green-from-green discrimination capability of the plant discrimination system.
The outdoor performance of the advanced real-time plant discrimination system has been evaluated, using three different plant species, Anthurium, Sunkisses and Dandelion. Experimental results have demonstrated that the system can attain reliable plant detection and practical discrimination at a vehicle speed of 3 km/h.
Future work is focusing on (a) achieving high detection accuracy at a target vehicle speed of 10 km/h, by addressing known system limitations, (b) further researching the concept of aggregating and thresholding plant NDVI measurements within defined regions-of-classification with the objective of statistically based automated threshold determination, for making single plant strike decisions, and (c) the statistical determination of optimal regions-of-classification for calibrating the PDU in-situ, prior to use.
5. ConclusionThe performance of an advanced real-time plant discrimination system, employing two red laser diodes (635 nm, 685 nm) and one near-infrared laser diode (785 nm) in conjunction with a linear sensor array, has been investigated experimentally. Experimental results have shown that the use of dual Normalised Difference Vegetation Indices (NDVI) in conjunction with data aggregation and threshold optimisation significantly enhances the green-from-green discrimination capability of the plant discrimination system.The outdoor performance of the advanced real-time plant discrimination system has been evaluated, using three different plant species, Anthurium, Sunkisses and Dandelion. Experimental results have demonstrated that the system can attain reliable plant detection and practical discrimination at a vehicle speed of 3 km/h.Future work is focusing on (a) achieving high detection accuracy at a target vehicle speed of 10 km/h, by addressing known system limitations, (b) further researching the concept of aggregating and thresholding plant NDVI measurements within defined regions-of-classification with the objective of statistically based automated threshold determination, for making single plant strike decisions, and (c) the statistical determination of optimal regions-of-classification for calibrating the PDU in-situ, prior to use.
การแปล กรุณารอสักครู่..
