Resumen:
This paper intends to implement registration of
images to temporal sequences of maize crops digitally sensed in
field trials through the use of a robotic platform. The
navigation of trials of this platform equipped with visible and
multispectral light sensors, and the registration processes used
allow us to generate “stacks images” by different genetic
varieties of maize with the aim of obtaining a phenological
characterization of them, and to compare them. In addition to
the basic problems of displacement, lighting, and angle of the
different captures, there is the drawback that the objects
(plants) are not the same and in many cases also not similar to
the previous images (different phenological stages). Added to
this is the complexity of sensing the images in outdoor
environments, and in particular the general conditions in the
field (uneven path surface, climatic conditions). The proposed
algorithms are part of the generation of multilayer digital
image data banks that allow, through other digital analysis and
processing techniques, the automated identification and
interpretation of different characteristics for the determination
of the response of the different genetic varieties of corn that is
studied.
