Enhancing Perceptibility of Barely Perceptible Targets
Abstract
A major virtue of our enhancement procedures is that targets do not have to be restricted to a fixed location nor does the exact location have to be known or discovered beforehand. This is obvious in the case of temporal manipulations, but it applies in general to our spatial manipulations as well. Contours that make an image region appear three-dimensional or that divide an image into figure and ground can be placed within a general area, or moved from region to region as the task requires. We found a similar lack of constraint on the types of images amenable to our enhancement procedures. Our procedures worked with such dissimilar images as randomly placed dots and short vectors, synthetic aperture radar images, small diagonal line segments, fragmented forms, and digitized photographs of faces, roadways, tanks and trucks obscured by various types of noise. Our work on this contract suggests that we can develop a large, varied, and flexible stockpile of image enhancement techniques. There are many different ways of imparting visual meaning to an image. There are many different image types amenable to enhancement. Our findings may prove applicable to such disparate situations as picking up key words or sentences in documents, spotting faces in a crowd, identifying objects in aerial reconnaissance, surveillance and badly damaged photographs, recognizing marginal video and facsimile transmissions, and interpreting camouflaged images.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 28, 1982
- Accession Number
- ADA127569
Entities
People
- Naomi Weisstein
Organizations
- University at Buffalo