A software evaluation methodology has been developed at SRI International for evaluating contributions to the ARPA/DMA Image Understanding Testbed. This paper describes the criteria that have shaped the evaluation methodology.
Computer vision publications
Image-To-Image Correspondence: Linear-Structure Matching
We examine the task of matching images of a scene when they are taken from very different vantage points, when there is considerable scale change, and when the image orientations are unknown.
Fractal-Based Description Of Natural Scenes
This paper addresses the problems of: representing natural shapes such as mountains, trees, and clouds, and computing their description from image data.
The DARPA / DMA Image Understanding Testbed Programmer’s Manual
The primary purpose of the Image Understanding (IU) Testbed is to provide a means for transferring technology from the DARPA-sponsored IU research program to DMA and other organizations in the defense community.
Choosing A Basis For Perceptual Space
If it is possible to interpret an image as a projection of rectangular forms, there is a strong tendency for people to do so. In effect, a mathematical basis for a vector space appropriate to the world, rather than to the image, is selected.
The Relationship Between Image Irradiance and Surface Orientation
A formulation of shape from shading is presented in which surface orientation is related to image irradiance without requiring detailed knowledge of either the scene illumination or the albedo of the surface material.
Shape From Shading: An Assessment
In dealing with the question of what surface parameters can be recovered locally from image shading, we show that, at most, shading determines relative surface curvature, i.e., the ratio of surface curvature measured in orthogonal image directions.
A General Approach To Machine Perception Of Linear, Structure In Imaged Data
In this paper we address a basic problem in machine perception: the tracing of “line-like” structures appearing in an image. It is shown that this problem can profitably be viewed as the process of finding skeletons in a gray scale image.
The Phoenix Image Segmentation System: Description and Evaluation
This report summarizes applications for PHOENIX, the history and nature of the algorithm, details of the Testbed implementation, the manner in which it is invoked and controlled, the type of results that can be expected, and suggestions for further development.