Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Task Specific Information Amit Ashok1, Pawan K. Baheti1 and Mark A. Neifeld1,2

Summary: Task Specific Information
Amit Ashok1, Pawan K. Baheti1 and Mark A. Neifeld1,2
Department of Electrical and Computer Engineering1
College of Optical Sciences2
1230 E. Speedway Blvd., University of Arizona, Tucson, AZ 85721
Email: {ashoka,baheti,neifeld}@ece.arizona.edu
Abstract: We introduce the notion of task-specific information (TSI) to quantify the performance
of imaging systems. We demonstrate the utility of TSI for evaluating the performance of conven-
tional and projective imagers for a detection task.
c 2007 Optical Society of America
OCIS codes: 110.3000,110.4280,200.3050.
1. Introduction
The information content of an image plays an important role in a wide array of applications ranging from video
compression to imaging system design[1,2]. However, the computation of image information content remains a chal-
lenging problem. For computational tasks such as automated target recognition, it is important to note that not all
information contained in the imagery is relevant to the task. For example, in a target detection task the final outcome
is a binary variable, representing the two target states "target present" or "target absent". Therefore, we could say
that the image may contain no more than 1 bit of relevant information for the target detection task. We will refer
to this relevant information as task-specific information (TSI). In this paper, we describe a formal approach to the
computation of TSI and use it to describe/quantify the task-specific performance of conventional and projective


Source: Ashok, Amit - Department of Electrical and Computer Engineering, University of Arizona


Collections: Engineering