Validation Of High Dynamic Range Imaging To
Luminance Measurement
Smita Anaokar, and Martin Moeck, Ph.D.
Abstract—High dynamic range imaging is a set of techniques that
allows for a far greater dynamic range of exposure than normal digital
imaging techniques. The intention is to accurately represent the
dynamic range of lighting levels found in real scenes, ranging from
direct sunlight to deep shadows. Various software packages have
emerged that translate these HDR images into high resolution
luminance maps. This paper tests the accuracy of one such package,
taking into consideration different Munsell hues, values and chroma. It
investigates the impact of different light spectra, spatial frequency,
vignetting, and thermal noise on the accuracy of luminance
measurements and determines the potential errors.
1 INTRODUCTION
High Dynamic Range (HDR) imaging was initially developed for photogra-phers. Since any single photograph does not accurately represent the
dynamic range of the actual scene, a series of images are taken from the darkest
image to the brightest image and are then combined to create one single HDR
image. This image contains the true dynamic range of the scene. HDR image
formats are especially significant for archival and stock uses, since they store
data with enough precision to record what was seen rather than displayed.
However, HDR imaging was not developed specifically for lighting measure-
ments, such as luminance or bi-directional reflectance distribution functions
(BRDF). Nevertheless, algorithms and software have been developed specifically
for this purpose.(Ward 2001; Seetzen, Heidrich and others, 2003) This paper
verifies the accuracy of this approach, taking into consideration Munsell hue,
value and chroma, lamp spectra, vignetting, and spatial resolution.
HDR imaging allows the use of consumer grade digital cameras. Their advan-
tages are as follows:
Department of Architectural Engineering, The Pennsylvania State University, State College,
PA-16801, USA
List of Symbols and Abbreviations: L, Luminance; E, Illuminance;
, Reflectance; , Pi 
3.14159265; HDR, High dynamic range; CRI, Color rendering index; CCT, Correlated color
temperature; WB White balance
L E U K O S V O L 2 N O 2 O C T O B E R 2 0 0 5 P A G E S 1 3 3 – 1 4 4
©2005 The Illuminating Engineering Society of North America
doi: 10.1582/LEUKOS.2005.02.02.005
133

• Resolution is high as compared to earlier research grade luminance CCD
cameras with photopic filters.
• Errors in luminance calibrated cameras can range up to 100 percent and
higher depending on factors such as vignetting, resolution and temperature.
• Consumer grade cameras have reliable built-in thermal noise reduction.
• There are a wide range of lenses available.
• The price of the camera as compared to luminance meters and luminance
calibrated cameras is very low.
The objective of this study was to verify the accuracy of consumer grade digital
cameras and HDR imaging.(Debevec and Malik 1997; Lewin and O’Farrell 1999;
Marschner, Lafortune and others, 1999; Andersen, Michel and others, 2001)
The software used to produce images in this study was Photosphere.(Inanici and
Galvin 2004) The Nikon Coolpix 5400 was selected for the experiments after
comparing the errors with that of an Olympus E-01 and a Canon Digital
Rebel.(Berry 1992; Bruno and Dominique 2003) Errors in the luminance
mapping were determined due to the following:
• Color of Munsell chips.
• Spectral power distribution of the source.
• Optical vignetting of the camera.
• Spatial resolution of the object.
2 ERROR DEPENDENCE ON COLOR
The error dependence on color was verified with the following set-up. Six Munsell
cards (gray, red, yellow, green, blue, purple) were used, each card consisting of
14 chips (16 chips for gray) of a specific hue (color) with varying Munsell values
(brightness) and chroma (saturation) as shown in Fig. 1. The size of each Munsell
chip was 1.5 2.3 cm, and the reflectance of each Munsell chip was specified for
different lamps and daylight spectra.
The experiment was done using the set up in Fig. 2.
The matte Munsell charts with individual Munsell chips were placed on a
horizontal surface with illuminance over the samples varying as shown in Fig. 2.
An N6 matte Munsell gray card with 30 percent reflectance was also placed in
the set up. Two sets of conditions were used, one with a high illuminance of 478
lx and the other with a low illuminance of 88 lx. The measurement conditions are
shown in Table 1.
The shutter speed of the camera was changed in one step increments to obtain
a series of twelve images such that a white pixel in the darkest image was not
completely white and a black pixel in the brightest image was not completely
black. These images were then used in Photosphere to create a HDR image. The
illuminance on the N6 gray card with a 30 percent reflectance was measured
during the experiment and the luminance of the gray card was computed using
the equation
L

E


(1)
The HDR image was calibrated using this luminance value. The HDR image
generated was used to obtain the luminance values which were then used to
compute the reflectance for each sample. The variation in the illuminance values
across the length of the Munsell card was measured as in Fig. 2, and a correction
L E U K O S V O L 2 N O 2 O C T O B E R 2 0 0 5 P A G E S 1 3 3 – 1 4 4
134