EMPA Media Technology

Publications 2012

Choice-Based Experiments in Multiple Dimensions

M. Scheller Lichtenauer, I. Sprow, P. Zolliker

Color technology needs specifications to which extent physical differences of stimuli correspond to differences in perception. Generalized linear models (GLMs) have proved successful to provide such specifications from choice-based experiments. However, the use of GLMs imposes practical restrictions on the experiment and stimulus parameters. We propose an alternative analytic approach based on machine learning and demonstrate its use in designing and analyzing choice-based experiments with multiple stimulus dimensions.

Color Research & Application, Volume 38, Issue 5, pages 334-343, 2013


Illustration of local contrast changes in gamut mapping. Picture: Iris Sprow

The Impact of Image-Difference Features on Perceived Image Differences

J. Preiss, I. Lissner, P. Urban, M. Scheller Lichtenauer, P. Zolliker

We discuss a few selected hypotheses on how the visual system judges differences of color images. We then derive five image-difference features from these hypotheses and address their relation to the visual processing. Three models are proposed to combine these features for the prediction of perceived image differences. The parameters of the image-difference features are optimized on human image-difference assessments. For each model, we investigate the impact of individual features on the overall prediction performance. If chromatic features are combined with lightness-based features, the prediction accuracy on a test dataset is significantly higher than that of the SSIM index, which only operates on the achromatic component.

CGIV 2012 Final Program and Proceedings

bibtex PDF Database
image difference map

Learning Image Similarity Measures from Choice Data

M. Scheller Lichtenauer, P. Zolliker, I. Lissner, J. Preiss, P. Urban

We present a corpus of experimental data from psychometric studies on gamut mapping and demonstrate its use to develop image similarity measures. We investigate whether similarity measures based on luminance (SSIM) can be improved when features based on chroma and hue are added. Image similarity measures can be applied to automatically select a good image from a sample of transformed images.

CGIV 2012 Final Program and Proceedings

bibtex PDF Database
two colour images mapping to the same grayscale imageimages derived from: Paul Klee. polyphon gefasstes Weiss, 1930, 140, Feder und Aquarell auf Papier auf Karton, 33,3 x 24,5 cm, Zentrum Paul Klee, Bern.

Image-Difference Prediction:From Grayscale to Color

I. Lissner, J. Preiss, P. Urban, M. Scheller Lichtenauer, P. Zolliker,

Existing image-difference measures show excellent accuracy in predicting distortions such as lossy compression, noise, and blur. Their performance on certain other distortions has room for improvement; one example is gamut mapping. This is partly because they do not interpret chromatic information correctly or ignore it entirely. We present an image-difference framework that comprises image normalization, feature extraction, and feature combination. Based on this framework, we create imagedifference measures by selecting specific implementations for each of the steps. Particular emphasis is placed on using color information to improve the assessment of gamut-mapped images. Our best image difference measure shows significantly higher prediction accuracy on a gamut-mapping dataset than all other evaluated measures.

IEEE Transactions on image processing accepted 15.08.2012

Example of an image difference mapExample of an image difference map

Image similarity for chromatic content

M. Scheller Lichtenauer, J. Preiss, P. Urban, P. Zolliker,

Image similarity describes the quality of an image relative to a reference. Image difference measures are calculated in order to quantify the perceived difference. The concept is also applicable to model where two colour images differ. Existing measures for image similarity focus on high spatial frequency distortions, best visible in the luminance channel. Lower frequency distortions resulting from gamut mapping, tone mapping or illuminance change are only incorporated as far as they influence luminance. As luminance reduction and chroma reduction are often correlated in real applications, we studied image similarity on artificial transformations, thus allowing us to reduce luminance or chrominance independently of each other.

Workshop Farbbildverarbeitung 2012

Example of a distorted image

Near Surface Swimming of Salmonella Typhimurium Explains Target-Site Selection and Cooperative Invasion

Benjamin Misselwitz, Naomi Barrett, Saskia Kreibich, Pascale Vonaesch, Daniel Andritschke, Samuel Rout, Kerstin Weidner, Milos Sormaz, Pascal Songhet, Peter Horvath, Mamta Chabria, Viola Vogel, Doris M. Spori, Patrick Jenny, Wolf-Dietrich Hardt

Targeting of permissive entry sites is crucial for bacterial infection. The targeting mechanisms are incompletely understood. We have analyzed target-site selection by S. Typhimurium. This enteropathogenic bacterium employs adhesins (e.g. fim) and the type III secretion system 1 (TTSS-1) for host cell binding, the triggering of ruffles and invasion. Typically, S. Typhimurium invasion is focused on a subset of cells and multiple bacteria invade via the same ruffle. It has remained unclear how this is achieved. We have studied target-site selection in tissue culture by time lapse microscopy, movement pattern analysis and modeling. Flagellar motility (but not chemotaxis) was required for reaching the host cell surface in vitro. Subsequently, physical forces trapped the pathogen for ~1.5-3 s in near surface swimming. This increased the local pathogen density and facilitated scanning of the host surface topology. We observed transient TTSS-1 and fim-independent stopping and irreversible TTSS-1-mediated docking, in particular at sites of prominent topology, i.e. the base of rounded-up cells and membrane ruffles. Our data indicate that target site selection and the cooperative infection of membrane ruffles are attributable to near surface swimming. This mechanism might be of general importance for understanding infection by flagellated bacteria.

PLoS pathogens