Abstract
When using a mobile camera to search for a target object, it is often important to maximize the efficiency of the search. We consider a method for increasing efficiency by searching only those subregions that are especially likely to contain the object. These subregions are identified via spatial relationships. Searches that use this method repeatedly find an “intermediate” object that commonly participates in a spatial relationship with the target object, and then look for the target in the restricted region specified by this relationship. Intuitively, such searches, calledindirect searches, seem likely to provide efficiency increases when the intermediate objects can be recognized at low resolutions and hence can be found with little extra overhead, and when they significantly restrict the area that must be searched for the target. But what is the magnitude of this increase, and upon what other factors does efficiency depend? Although the idea of exploiting spatial relationships has been used in vision systems before, few have quantitatively examined these questions.
We present a mathematical model of search efficiency that identifies the factors affecting efficiency and can be used to predict their effects. The model predicts that, in typical situations, indirect search provides up to an 8-fold increase in efficiency. Besides being useful as an analysis tool, the model is also suitable for use in an online system for selecting intermediate objects.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Ahlswede, R., and Wegener, I. 1987.Search Problems, Wiley: New York.
Aloimonos, J. 1990. Purposive and qualitative active vision,AAAI Qualitative Vision Workshop, pp. 1–5.
Bajcsy, R. 1988. Active perception,Proc. IEEE, 76: 996–1005, August.
Ballard, D.H., and Brown, C.M. 1982.Computer Vision, Prentice-Hall: Englewood Cliffs, NJ.
Ballard, D.H., and Brown, C.M. 1992. Principles of animate vision,Comput. Vis., Graph., Image Process., 56(1): 3–21.
Barrow, H.G., and Tenenbaum, J.M. 1976. MSYS: a system for reasoning about scenes, Technical Note 121, AI Center, SRI International, March.
Bolle, R.M., Califano, A., and Kjeldsen, R. 1989. Data and model driven foveation, Research report, Exploratory Computer Vision Group, IBM T.J. Watson Research Center.
Bolles, R.C 1977. Verification vision for programmable assembly,Proc. 5th Intern. Joint Conf. Artif. Intell., Cambridge, MA.
Burt, P.J. 1988. Smart sensing within a pyramid vision machine,Proc. IEEE, 76: 1006–1015, August.
Garey, M.R., and Johnson, D.S. 1979.Computers and Intractability: A Guide to the Theory of NP-Completeness, Freeman: New York.
Garvey, T.D. 1976. Perceptual strategies for purposive vision, Technical Note 117, SRI International, September.
Johnson, D.T., and Schubert, L.K. 1982. A planning control strategy that allows for the cost of planning,6th European Meeting on Cybernetics and Systems Research, April.
Larsen, R.J., and Marx, M.L. 1981.An Introduction to Mathematical Statistics and its Applications, Prentice-Hall: Englewood Cliffs, NJ.
Maver, J., and Bajcsy, R. 1993. Occlusions as a guide for planning the next view,IEEE Trans. Patt. Anal. Mach. Intell., 15: 417–433, May.
McKeown, Jr., D.M., Harvey, Jr., W.A., and McDermott, J. 1985. Rule-based interpretation of aerial imagery,IEEE Trans. Patt. Anal. Mach. Intell., 7: 570–585, September.
Reece, D.A. 1992. Selective perception for robot driving, Tech. Rept. CMU-CS-92-139, Carnegie Mellon Computer Science, May.
Reece, D.A., and Shafer, S. 1991. Using active vision to simplify perception for robot driving, Tech. Rept. CMU-CS-91-199, Carnegie Mellon Computer Science, November.
Rimey, R.D., and Brown, C.M. 1992. Where to look next using a Bayes net: Incorporating geometric relations,Proc. 2nd Europ. Conf. Comput. Vis., Ligure, Italy.
Rimey, R.D., and Brown, C.M. 1993. Control of selective perception using Bayes nets and decision theory,Intern. Comput. Vis., this issue.
Russell, D.M. 1978. Constraint networks: modeling and inferring object locations by constraints, Tech. Rept. 38, University of Rochester Computer Science Dept., August.
Sarachik, K.B., and Grimson, W.E.L. 1993. Gaussian error models for object recognition,Proc. Conf. Comput. Vis. Patt. Recog., June.
Swain, M.J. 1990. Color indexing, Tech. Rept. 360, University of Rochester Computer Science Dept.
Swain, M.J., Kahn, R.E., and Ballard, D.H. 1992. Low resolution cues for guiding saccadic eye movements,Proc. IEEE Conf. Comput. Vis. Patt. Recog., Urbana Champaign, IL, June.
Tarabanis, K., Tsai, R.Y., and Allen, P.K. 1992. The MVP sensor planning system for robotic vision tasks, Tech. Rept., Columbia University Computer Science Department.
Tsotsos, J.K. 1992. Active vs. passive visual search: Which is more efficient?,Intern. J. Comp. Vis., 7: 2.
Van Trees, H.L. 1968.Detection, Estimation, and Modulation Theory, vol. 1, Wiley: New York.
Wilkes, D., and Tsotsos, J.K. 1992. Active object recognition,Proc. IEEE Conf. Comput. Vis. Patt. Recog., Urbana Champaign, June.
Wixson, L.E. 1992. Exploiting World Structure to Efficiently Search for Objects, Tech. Rept. 434, University of Rochester Computer Science Department, July.
Wixson, L.E. 1994.Searching for Objects in 3D Space, Ph.D. thesis, University of Rochester Computer Science Dept., forthcoming.
Wixson, L.E., and Ballard, D.H. 1989. Real-time detection of multicolored objects,SPIE Sensor Fusion II: Human and Machine Strategies, vol. 1198, November.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Wixson, L.E., Ballard, D.H. Using intermediate objects to improve the efficiency of visual search. Int J Comput Vision 12, 209–230 (1994). https://doi.org/10.1007/BF01421203
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF01421203