Ethan D. Montag, Ph.D.
Assistant Professor

Ph.D., Experimental Psychology
University of California, San Diego

1065 Color Science Building - Bldg. 18
(585)475-5096
E-Mail: montag@cis.rit.edu


Professional Experience

2000 to Present Assistant Professor, Munsell Color Science Laboratory, Chester F. Carlson Center for Imaging Science

1996 to 2000 Research Assistant Professor, Munsell Color Science Laboratory.

1994 to 1996 Post-Doctoral Fellow, Munsell Color Science Laboratory.

1991 to 1994 Post Doctoral Fellow, Center for Visual Science, University of Rochester.

1985 to 1991 Teaching and Research Assistant, Department of Psychology,
University of California, San Diego.


Vita


Membership of Professional Bodies

Optical Society of America
The Society for Imaging Science and Technology
Inter-Society Color Council
International Colour Vision Society


Research and Teaching

Visualization

 
My interests in visualization involve both theoretical and applied aspects of using images to convey information to users.

On the theoretical side, many imaging applications involve the visualization of multidimensional data. Because color space is multidimensional and many different paths can be defined within color space, there is a potential that color can be used to convey this information. I have been involved in research to evaluate the use of color in conveying information by measuring human performance in reading images.

Unidimensional Data

The two images below convey digital elevation data from the US Geological Survey. The image on the left shows a 3D rendering of the data that is pretty but arguably useless for actually visually determining data values. On the right the same data is encoded using a lightness scale that is perceptually linear (on a calibrated monitor). In addition, hue information is added which allows users to efficiently utilize a legend to interpret the image.

 

See:
Montag, E. D. (1999) The Use of Color in Multidimensional Graphical Information Display. IS&T/SID Seventh Color Imaging Conference, Scottsdale. Download the pdf.

High-Dynamic Range Imaging

Conventional displays can only display 8-bits of information while modern sensors can record a much higher bit depths. We have borrowed the techniques used for the rendering of pictorial HDR images and evaluated them for a variety of scientific imagery.

In the example above, we show an MR scan rendered linearly, on the left, and rendered using eight different methods.

See:
Park, S. H. & Montag, E. D. (2004) Rendering Non-Pictorial (Scientific) High Dynamic Range Images, IS&T/SID Twelfth Color Imaging Conference, Scottsdale. Download the pdf.

Park, S. H. and Montag, E. D. Evaluating Tone Mapping Algorithms for Rendering Non-Pictorial (Scientific) High-Dynamic-Range Images, (submitted to ACM: Transactions on Applied Perception). Download the pdf.

Multidimensional Data

We are also exploring the use of color in the visualization of multi- and hyperspectral data. My PhD student, Hongqin Zhang, has been looking at methods for the perceptual rendering of remotely sensed (i.e., from aircraft and satellites) spectral images based on ICA and PCA using perceptually based color spaces. She is also exploring the use of color for encoding abundance maps.

   

Image Quality

 

Traditionally, investigation of image quality involves the a priori identification of factors involved in the perception of images and the manipulation of these factors to investigate there effect on image quality. We are taking a different approach in our research. By using multidimensional analysis of data collected in psychophysical evaluation of images we are able to correlate psychological factors involved in image quality judgments with physical parameters of the image. For example, we found that subjects based there judgments of image preference for a series of portraits made on different printers on the model's skin tone. Using our data we can predict the optimal color balance for the skin tone. The picture below shows an example of how color balance in CIELAB space affects the appearance of a portrait.

By controlling and manipulating aspects of the output, we hope to be able to quantify and predict the factors involved in image quality.

   


Color Gamut Mapping

 

The spinning image on this page is a representation of the gamut of a color CRT in CIELAB space. In my research on gamut mapping this gamut is the original gamut of an original color image device.

When an image is reproduced on a another device, the range of colors available may be different from that in the original gamut. If the second device has a smaller range of colors, the image must be manipulated in some way to produce a veridical representation.

Our past research efforts have concentrated on limiting the gamut in one color dimension at a time (for example, chroma or lightness) and then examining gamut mapping techniques for simple rendered images consisting of individual spheres of one color. Our current research is extending the findings of the previous research by using more "realistic" gamuts with more complex images.

See:
Montag, E. D. and Fairchild, M. D. (1997) Psychophysical evaluation of gamut mapping techniques using simple rendered images and artificial gamut boundaries. IEEE Transactions on Image Processing, 6, 977-989. pdf

Montag, E. D. and Fairchild, M. D. (1998) Gamut mapping: Evaluation of chroma clipping techniques for three destination gamuts. IS&T/SID Sixth Color Imaging Conference, Scottsdale, 57-61. Download the pdf.

rotating monitor gamut
Color gamut of a CRT

 

 

 

 

 

 

Click here to look inside a gamut.

   

Industrial Color Tolerance

 

MCSL has a long history in research involving industrial color tolerance both in the psychophysical collection of color difference data and the modeling of color difference equations.

I have been involved in research testing the feasibility of using CRTs to simulate color differences of hard copy and the psychophysical measurement of color tolerances using uniform patches and simulated textures. By developing techniques to measure color tolerance on a CRT we hope to speed up data acquisition and perhaps use rendering techniques to simulate the appearance of surfaces for determining color tolerances on different media (textiles, auto paint, etc.).

The figure on the left shows color tolerances for uniform patches (triangles) and textures (squares and circles) plotted as traditional threshold-vs-intensity (TVI) plots on log-log axes. A slope of 1 would indicate Weber's law. This representation is much less complex than the plot of the tolerances in CIELAB L*.

See:
Montag, E. D. and Berns, R. S., Lightness Dependencies and the Effect of Texture on Suprathreshold Lightness Tolerance, Color Res. and Appl. 25, 241-249, (2000). Download the pdf.

   

Teaching
 

I currently teach the following courses:

Vision & Psychophysics 1050-701,1051-774
Color Science Seminar 1050-801
Applied Colorimetry Lab 1050-712

Vision & Psychophysics is also taught as a Distance Learning course. Please click here to find out more about the distance learning MS in Imaging Science.

My distance-learning course, Vision & Psychophysics, runs concurrently with the local version of the class. Because the distance students are unable to attend lecture, I have created a web site which takes the place of the lectures. If you are interested in seeing how this works please click here.

 

 


Here's a picture illustrating the benefits of color in imaging.