Automatic mosaicking of very spatial resolution UAV multispectral images for precision agriculture: test of MICMAC freeware
Résumé
For precision agriculture purposes, use of UAV could be helpful to create on demand centimeter resolution multispectral ortho-images of a plot. This high resolution imagery can then be used for instance in weed detection and treatment. However to be efficient, the computation chain between images acquired from UAV and ortho-rectified useful data should be fully automatic, fast and cost efficient. This paper evaluates the ability of “MICMAC” freeware to support such process. It focuses particularly on geometric precision aspects when using multiple sensors (visible and near infrared, NIR). Tests were conducted on an experimental plot in Spain using visible and NIR sensors. It concludes on good results when using only one visible RGB camera and the standard MICMAC processing chain. Independent processing with MICMAC of NIR images does not lead to enough accurate geometric precision. Further developments are suggested to tackle this issue.