Dr. Ruth Ellen Rosenholtz

Principal Research Scientist

Primary DLC

Department of Brain and Cognitive Sciences

MIT Room: 32-D426

Areas of Interest and Expertise

Computational Modeling of Human Vision, Particularly Visual Search/Attention and Texture Perception
Application of Understanding of Human Vision to Design of User Interfaces and Information Visualizations
Clutter

Research Summary

Dr. Ruth Rosenholtz studies human vision, including visual search, peripheral vision, perceptual organization, and the impact of visual clutter on task performance. We take a three-pronged approach: (1) Computational modeling (computer vision-based, ideal observers, Monte Carlo simulations, and neurobiologically-inspired); (2) Behavioral experiments; and (3) Applying our models and understanding of human vision to applications such as image compression, design of user interfaces, and design of information visualizations.

RESEARCH INTERESTS:

The visual system as statistician
(*) Does the visual system collect summary statistics in early vision?
(*) A model of peripheral vision
(*) Predicting recognition in the periphery
(*) Predicting visual search performance
(*) Predicting perception of visual illusions

Texture perception
(*) Shape-from-texture
(*) Models for texture segmentation.
(*) Segmenting images into textured and non-textured regions.

Visual search
(*) Visual search in cluttered environments – what is clutter?
(*) Models for visual search and “popout.”
(*) “Asymmetries” in visual search
(*) Effects of background color on color search

Application of human vision research to user interface design and information visualization
(*) Predicting what groups people will perceive in a design.
(*) Searching the web with enhanced thumbnails.
(*) Document browsing aids.
(*) “Doodle” icons to aid in searching for a computer file.
(*) Tools for visualizing a large document on a small display.
(*) Understanding clutter.

Perceptually-based image compression and image quality
(*) Reducing blocking effects in block transform coded images.
(*) Perceptually based coding of still images.

Recent Work