Multiple sensor-based weed segmentation

Publication Type:
Journal Article
Citation:
Proceedings of the Institution of Mechanical Engineers. Part I: Journal of Systems and Control Engineering, 2010, 224 (7), pp. 799 - 810
Issue Date:
2010-11-01
Metrics:
Full metadata record
Files in This Item:
Filename Description Size
Thumbnail2009008418.pdf673.85 kB
Adobe PDF
Bidens pilosa L (commonly known as cobbler's peg) is an annual broad-leaf weed in tropical and subtropical regions and reportedly needs to be identified and eliminated when farming 31 different crop varieties. This paper presents a multi-modal sensing approach for detecting Bidens leaves within wheat plants. Visual cue-based automatic discrimination of Bidens and wheat leaves is non-trivial owing to the curled-up nature of the wheat leaves. Therefore, spectral responses of Bidens and wheat leaves are first analysed to understand the discriminative spectral bands. Then a multi-modal sensory system consisting of a near infra red (NIR) and a visual camera set-up is proposed. Information retrieved from the sensory set up is then processed to generate a series of cues that are fed into a classification algorithm. Classification results are validated through experimentation. The proposed technique is able to achieve an accuracy of 88-95 per cent even when there is substantial overlapping between Bidens and wheat leaves. Further, it is also shown that the algorithm is robust enough to discriminate some other commonly available plant species.
Please use this identifier to cite or link to this item: