Applied Statistics Helps in Extracting Information
Exhibit Code: GLE-124
Zone: Engineering Park
Location: James E. Gleason Hall (GLE/009) - Next to Toyota Lab
Time: All Day
In a hyperspectral image, each pixel is represented by a spectral curve showing how much light is observed both in the visible range as well as in a wider range of (invisible) electromagnetic waves. In practice, the spectral curve is digitized into hundreds of narrow spectral bands and can be described using a set of, letís say, 200 numbers. This is in contrast to a traditional color picture that you see on a computer screen or in print, in which each pixel is represented by an intensity of three (or four) basic colors, and the pixel information (color) can be represented using only three or four numbers. Considering that a hyperspectral image may consist of hundreds of thousands of pixels, there is a lot of data to analyze. Statistics, an indispensable tool in such situations, allows us to extract the most important information from the data and to build predictive statistical models. Visitors will learn about current research in Center for Quality and Applied Statistics related to how statistics helps in obtaining important information contained in images. We will show how very small objects (smaller than the pixel size) can be detected in some images.
Peter Bajorski, Ernest Fokoue, Eric Spence , Jerry Stocker , Teresa Ludington , Jo Bill , Chieh-Chang Hsu , ZhiYan Lin , Thong Nguyen , Wing Liang , Randy Brown
43 people have added this exhibit