Spatial Wavelet Analysis:
A Per-Object Remote Sensing Method
Background*. Spatial Wavelet Analysis (SWA) is a powerful image-processing
technique that has considerable potential to quantify spatial landscape and
plant patterns at multiple scales and across large areas. SWA can be applied to any digital image, whether aerial photography,
Lidar, or Satellite data.
Wavelets have been used across a wide range of scientific disciplines from
medical imaging to astronomy, to identify the size, shape, and location of
individual features of interest. However, though wavelets hold tremendous
promise for objectively and automatically quantifying ecological features in
remotely sensed imagery, this application has gone largely unexplored. Our group is currently involved within several projects to demonstrate and
assess how Spatial Wavelet Analysis of aerial photography, Lidar data, and
satellite imagery can help provide spatial plant information. Accurate estimates
of spatial plant patterns across large areas will allow ecologists to infer
regarding relationships between landscape patterns and underlying ecological and
environmental processes.
* see the following references
for more detailed information:
Strand, E.K., Smith A.M.S., Bunting, S.C., Vierling, L.A.,
Hann, D.B. and Gessler, P.E., (2006) Wavelet estimation of
vegetation spatial patterns in multi-temporal aerial photography,
International Journal of Remote Sensing, 27, 9-10, 2049-2054. (PDF)
Falkowski, M.J., Smith, A.M.S., Hudak, A.T., Gessler, P.E., Vierling, L.A. and
Crookston, N.L., (2006). Automated estimation of individual conifer tree
height and crown diameter via Two-dimensional spatial wavelet analysis of lidar
data, Canadian Journal of Remote Sensing, 32, 1, 153-161 (PDF)
Evaluation of Individual Rangeland Plant Location and Crown Widths.
Our research group has performed analysis of
two quite different rangeland research objectives. Namely, the assessment of
western juniper woody encroachment and the evaluation of structurally
smaller shrubs such as: Big Sage Brush, Antelope Bitterbrush, and Stiff
Sage. In the above image, the dark points in the left image are the junipers. The green circles in the
right image represents the projected 'crown width' of each juniper. In a recent
study this method was also applied to experimental aerial photography
produced from the AeroCAM sensor.
The images to the right depicts 0.25m spatial
resolution AeroCAM data where the shrubs also appears as dark dots on a lighter
grass/forbs background.
Note that the method in both cases successfully
identifies the plants regardless of whether the surrounding matrix (i.e.
background) is dark or light. i.e. the method is ~ background invariant.
* see the following references
for more detailed information:
Garrity, S.R., Vierling, L.A., Smith, A.M.S. and Hann, D.B.,
Suitability of Spatial Wavelet Analysis (SWA) for the automatic assessment of
shrubs within an arid environment, Canadian Journal of Remote Sensing, in review.
Strand, E.K., Robinson, A.P. and Bunring, S.C. (2007)
Spatial patterns on the sagebrush steppe/Western juniper ectone, Plant Ecology,
190, 2, 159-173.
Strand, E.K., Smith A.M.S., Bunting, S.C., Vierling, L.A.,
Hann, D.B. and Gessler, P.E., (2006) Wavelet estimation of
vegetation spatial patterns in multi-temporal aerial photography,
International Journal of Remote Sensing, 27, 9-10, 2049-2054. (PDF)
Evaluation of Individual Forested Tree Location, Crown Widths, and
Structure. We are also researching the ability of SWA to evaluate the
size and structure of forested vegetation. The figure below depicts
various trees within an open-canopy mixed conifer forest stand in northern
Idaho.
In this case, SWA is used to identify light (tall height) objects on a dark (low
height) background.
* see the following reference for more detailed information:
Falkowski, M.J., Smith, A.M.S., Hudak, A.T., Gessler, P.E., Vierling, L.A. and
Crookston, N.L., (2006). Automated estimation of individual conifer tree
height and crown diameter via Two-dimensional spatial wavelet analysis of lidar
data, Canadian Journal of Remote Sensing, 32, 1, 153-161 (PDF)
Application of SWA to Spectral Data.
It is apparent that SWA like other object-oriented remote sensing methods
provides structural information that is quite different to
information that are normally
derived via spectral remote sensing data. One question that arises is: what
sort of information can we derive from identifying objects in spectral
imagery? Below is an example of an early version of SWA applied to a
temporal series of Landsat ETM data of center pivot irrigation crop circles.
By calculating the NDVI (or other index) value within the crop circle we can
chart greenness and perhaps water usage of these circles over time:
Last Updated May 21st 2007
|