Optical Image Analysis

The study field in the Matter Valley is characterized by several relatively fast moving rock glaciers. Quasi permanent GPS stations placed on selected rocks are used to frequently retrieve high accurate position estimates. As explained in the previous sections, these measurements are important to extend our understanding of the rock glacier behavior and its physical connection with the environment on a subdaily to daily resolution level.

Another important aspect is to study the displacement behavior in the spatial domain. Since resources other techniques rather than single point GPS measurements come into play. We use images - taken in the optical spectrum - from nearby installed cameras to extract displacements within the area that is covered by the sensor.

 

velocityField_mod
An example of Optical Flow estimation using an adaptive least-squares matching technique within the area of interest. This result was obtained using two image pairs in the August 2008 period, taken approximatelly 20 days apart. Arrow lengths are not in pixel scale with the image but show the relative displacement with respect to each other.

 

The rough surface topography turned out to be very suitable for estimating the displacement fields. The accuracy of the method and the projected displacement vectors depends on several factors:

  • Focal length of the optical system
  • Sensor resolution (pixel size)
  • Inherent algorithmic accuracy
  • Distance between camera and object of interest
  • Geometrical situation (projection parameters)

The two last points need to be evaluated for every vector position in the field of view of the camera. The varying distances between the camera and the local area of interest and the actually covered area (geometrical situation) may strongly vary, depending were on the image we apply the displacement detection.

Therefore further processing is necessary to accurately interpret these results.