MSc thesis subject: Tackling illumination dependence in Digital Hemispherical Photography

Digital Hemispherical Photography (DHP) is a technique to measure canopy Plant Area Index (PAI), which is traditionally used by foresters and ecologists to characterize (forest) ecosystems, as well as for ground truthing in biophysical parameter retrieval by optical remote sensing. DHP makes use of wide field of view (fish-eye) lenses and commercial user cameras (especially DSLR). The captured images are manually or automatically classified into sky and canopy elements to derive gap fraction (Pgap). Pgap together with assumptions of light propagation through the canopy is used to estimate PAI.

A major challenge in DHP is the correct parameter set-up (aperture/shutter time, ISO, metering method etc). Since photographic preferences are based on beauty perceptions by photo-artists, automatic choices usually don’t meet criteria for good separability of sky and canopy. Additionally, the dynamic range in upward pointing photo is typically higher than the range the sensor can capture. This makes DHP dependent on illumination conditions. The photoindustry established high dynamic range (HDR) photography to overcome large dynamic ranges for end-users. This method exploits the combined dynamic range of several pictures taken of the same scene, thereby virtually extending the dynamic range of the camera sensor. This technique has not yet been tested for scientific purposes.


  • Create data set for comparison (fieldwork in Speulderbos)
  • Literature review of how to deal with different illumination conditions [?]
  • Literature review on HDR photography background and algorithms
  • Investigate (open access) software tools for HDR compositing
  • Develop method(s) that exploits high dynamic range capabilities of camera system available in the GRS group
  • Compare time series/data set of PAI derived from HDR DHP with simultaneous measurement from illumination independent terrestrial lidar



  • Affinity for software tools
  • Scripting skills (e.g. R, Python, MatLab) are a preference

Theme(s): Sensing & measuring