Exploiting the generic viewpoint assumption

Abstract

Thegeneric viewpointässumption states that an observer is not in a special position relative to the scene. It is commonly used to disqualify scene interpretations that assume special viewpoints, following a binary decision that the viewpoint was either generic or accidental. In this paper, we apply Bayesian statistics to quantify the probability of a view, and so derive a useful tool to estimate scene parameters. This approach may increase the scope and accuracy of scene estimates. It applies to a range of vision problems. We show shape from shading examples, where we rank shapes or reflectance functions in cases where these are otherwise unknown. The rankings agree with the perceived values. Vision, 20(3), 243-261, 1996 This work may not be copied or reproduced in whole or in part for any commercial purpose. Permission to copy in whole or in part without payment of fee is granted for nonprofit educational and research purposes provided that all such whole or partial copies include the following: a notice that such copying is by permission of Mitsubishi Electric Research Laboratories, Inc.; an acknowledgment of the authors and individual contributions to the work; and all applicable portions of the copyright notice. Copying, reproduction, or republishing for any other purpose shall require a license with payment of fee to Mitsubishi Electric Research Laboratories, Inc. All rights reserved. The \generic viewpoint" assumption states that an observer is not in a special position relative to the scene. It is commonly used to disqualify scene interpretations that assume special viewpoints, following a binary decision that the viewpoint was either generic or accidental. In this paper, we apply Bayesian statistics to quantify the probability of a view, and so derive a useful tool to estimate scene parameters. Generic variables can include viewpoint, object orientation, and lighting position. By considering the image as a (dierentiable) function of these variables, we derive the probability that a set of scene parameters created a given image. This scene probability equation has three terms: the delity of the scene interpretation to the image data; the prior probability of the scene interpretation; and a new genericity term, which favors scenes likely to produce the observed image. The genericity term favors image interpretations for which the image is stable with respect to changes in the generic variables. It results from integration over the generic variables, using a low-noise approximation common in Bayesian statistics. International Journal of ComputerThis approach may increase the scope and accuracy of scene estimates. It applies to a range of vision problems. We show shape from shading examples, where we rank shapes or reectance functions in cases where these are otherwise unknown. The rankings agree with the perceived values.

Real-Time Feature Matching using Adaptive and Spatially Distributed Classification Trees

“…This is always the case in our study. In fact, research has been done regarding the generic view assumption and the rareness of accidental views [4].…”

Recognition times of different views of 56 depth-rotated objects: A note concerning Verfaillie and Boutsen (1995)

“…We are presently exploring the possibility ofquantifying the information objectively present in a given view (see, e.g., Freeman, 1994Freeman, , 1996Weinshall, Werman, & Tishby, 1996), in order to relate judged goodness and recognition times to objective image properties (the work ofCutzu & Edelman, 1994, shows that this does not necessarily preclude the potential of accounting for subject-specific, idiosyncratic factors), but this is beyond the scope of the present note. …”

Object Recognition by a Cascade of Edge Probes

“…The elevation of the camera varies by approximately 25 degrees with respect to the object, and the extent of scale variation across images is about 10%. For some of our images, the generic viewpoint assumption is violated [15]; for example two of the legs of the chair in Figure 4(a), second row, fourth column, are accidentally aligned.…”