Abstract
We advance new active computer vision algorithms based on the Feature space Trajectory (FST) representations of objects and a neural network processor for computation of distances in global feature space. Our algorithms classify rigid objects and estimate their pose from intensity images. They also indicate how to automatically reposition the sensor if the class or pose of an object is ambiguous from a given viewpoint and they incorporate data from multiple object views in the final object classification. An FST in a global eigenfeature space is used to represent 3D distorted views of an object. Assuming that an observed feature vector consists of Gaussian noise added to a point on the FST, we derive a probability density function for the observation conditioned on the class and pose of the object. Bayesian estimation and hypothesis testing theory are then used to derive approximations to the maximum a posterioriprobability pose estimate and the minimum probability of error classifier. Confidence measures for the class and pose estimates, derived using Bayes theory, determine when additional observations are required, as well as where the sensor should be positioned to provide the most useful information.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.References
Bajcsy R. Active perception. Proc IEEE 1988; 76: 996–1004.
Abbott AL, Ahuja N, Auen PK et al. Promising directions in active vision. Int J Computer Vision 1993; 11: 109–126.
Callari FG, Ferrie FP. Active recognition: using uncertainty to reduce ambiguity. Proc 13th Int Conf Pattern Recognition 1996; 1: 925–929.
Casasent D, Neiberg L. Classifier and shift-invariant ATR neural networks. Neural Networks 1995; 8(7/8): 1117–1130.
Stein F, Medioni G. Structural indexing: Efficient —D object recognition. IEEE Trans PAMI 1992; 14: 125–144.
Therrien CW. Decision, Estimation, and Classification: An Introduction to Pattern Recognition and Related Topics. New York: Wiley, 1989.
Sipe M, Casasent D, Neiberg L. Feature space trajectory representation for active vision. In Applications and Science of Artificial Neural Networks III, S. Rogers, (editor), Proc Soc Photo-Optical Instrumentation Eng 1997; 3077: 254–265.
Neiberg L, Casasent D, Fontana R, Cade J. Feature space trajectory neural net classifier: 8-class distortion-invariant tests. In Intelligent Robots and Computer Vision XIV: Algorithms, Techniques, Active Vision, and Materials Handling, D. Casasent (editor), Proc Soc of Photo-Optical Instrumentation Eng 1995; 2588: 540–555.
Casasent D, Shenoy R. Feature space trajectory for distorted-object classification and pose estimation in synthetic aperture radar. Opt Eng 1997; 36: 2719–2728.
Casasent D, Neiberg L. Time-sequential neural net classifier and pose estimator. Proc World Congress on Neural Networks 1996: 345–350.
Krumm J. Eigenfeatures for planar pose measurement of partially occluded objects. Proc IEEE Con Computer Vision and Pattern Recognition 1996: 55–60.
Baum EB, Haussler D. What size net gives valid generalization? Neural Computation 1989; 1: 151–160.
Smieja FJ, Richards GD. Hard learning the easy way: backpropagation with deformation. Complex Systems 1988; 2: 671–704.
Horn KSV, Martinez TR. The minimum feature set problem. Neural Networks 1994; 7(3): 491–494.
Denoeux T, Lengelle R. Initializing backpropagation networks with prototypes. Neural Networks 1993; 6(3): 351–363.
Powell MJD. Restart procedures for the conjugate gradient method. Math Programming 1977; 12: 241–254.
Casasent D, Barnard E. Adaptive-clustering optical neural net. Appl Opt 1990; 29: 2603–2615.
Raudys S, Jain A. Small sample size effects in statistical pattern recognition: Recommendations for practitioners. IEEE Trans PAMI 1991; 13: 252–264.
Jain A, Waller W. On the optimal number of features in the classification of multivariate Gaussian data. Patt Recognition 1978; 10: 365–374.
Neiberg L, Casasent D. Feature space trajectory neural network classifier. In Applications and Science of Artificial Neural Networks, S. Rogers and D. Ruck (editors), Proc Soc Photo-Optical Instrumentation Eng 1995; 2492: 361–372.
Rumelhart DE, Hinton GE, Williams RJ. Parallel Distributed Processing, Vol. 1. Cambridge, MA: MIT Press, 1986, pp. 318–362.
Poggio T, Girosi F. Networks for approximation and learning. Proc IEEE 1990; 78: 1481–1497.
Finnoff W, Hergert F, Zimmerman HG. Improving model selection by nonconvergent methods. Neural Networks 1993; 6(6): 771–783.
Musavi M, Ahmed W, Chan K, Faris K, Hummels D. On the training of radial basis functions. Neural Networks 1992; 5(4): 595–603.
Hebert M, Krotkov E. 3D measurements from imaging laser radars: how good are they? Image and Vision Computing 1992; 10: 170–178.
Wells G, Venaille C, Torras C. Vision-based robot positioning using neural networks. Image and Vision Computing 1996; 14: 715–732.
Dickinson S, Christensen H, Tsotsos J, Olofsson G. Active object recognition integrating attention and viewpoint control. Computer Vision and Image Understanding 1997; 67(3): 239–260.
Tarabanis K, Tsai R, Kaul A. Computing occlusion-free viewpoints. IEEE Trans PAMI 1996; 18: 279–292.
Murase H, Nayar SK. Visual learning and recognition of 3-D objects from appearance. Int J Computer Vision 1995; 14: 5–24.
Murase H, Nayar S. Detection of 3d objects in cluttered scenes using hierarchical eigenspace. Patt Recognition Lett 1997; 18(4): 375–384.
Leonardis A, Bischof H. Dealing with occlusions in the eigenspace approach. Proc IEEE Conf Computer Vision and Pattern Recognition 1996; 453–458.
Gonzalez RC. Syntactic Pattern Recognition. Reading, MA. Addison-Wesley, 1978.
Aeberhard S, Coomans D, Vel, OD. Comparative analysis of statistical pattern recognition methods in high dimensional settings. Patt Recognition 1992; 27: 1065–1077.
Manjunath BS, Chellappa R. Unsupervised texture segmentation using Markov random field models. IEEE Trans PAMI 1991; 13: 478–482.
Casasent D, Ashizawa S. Synthetic aperture radar detection, recognition, and clutter rejection with new minimum noise and correlation energy filters. Opt Eng 1997; 36: 2729–2736.
Rothman P. Syntactic pattern recognition. AI Expert 1992; 7: 40–51.
Hotelling H. Analysis of a complex of statistical variables into principal components. J Educ Psychology 1933; 24: 417–441, 498–520.
Swets DL, Weng JJ. Using discriminant eigenfeatures for image retrieval. IEEE Trans PAMI 1996; 18: 831.
Fukunaga K, Koontz WLG. Applications of the Karhunen-Loeve expansion to feature selection and ordering. IEEE Trans Comp 1970; C-19: 917–923.
Fisher RA. Contributions to Mathematical Statistics. New York: Wiley, 1950.
Talukder A, Casasent D. A general methodology for simultaneous representation and discrimination of multiple object classes. Opt Eng 1998; 37: 904–913.
Talukder A, Casasent D. Classification and pose estimation for active vision using non-linear features. Proc Soc Photo-Optical Instrumentation Eng 1998; 3386.
Duda RO, Hart PE. Pattern Classification and Scene Analysis. New York: Wiley, 1973.
Specht DF. Probabilistic neural networks. Neural Networks 1990; 3(1).
Nene SA, Nayar SK, Murase H. Columbia image object library (COIL-20). Technical Report CUCS-006-96, Department of Computer Science, Columbia University, New York, NY 10027, 1996.
Papoulis A. Probability, Random Variables, and Stochastic Processes. New York: McGraw-Hill, 1984.
Casasent D, Sipe MA. Feature space trajectory representation and processing for active vision. In Intelligent Robots and Computer Vision XV: Algorithms, Techniques, Active Vision, and Materials Handling: Proc Soc Photo-Optical Instrumentation Eng 1996; 2904: 390–401.
Regli WC, Gaines DM. National repository for design and process planning. National Science Foundation Design and Manufacturing Conference, Seattle, WA, January 1997.
Foley JD, van Dam A, Feiner SK, Hughes JF. Computer Graphics: Principles and Practice. Reading, MA: Addison-Wesley, 1996.
Van Trees HL. Detection, Estimation, and Modulation Theory. New York: Wiley, 1968, pp. 19–165.
Johnson R. Elementary Statistics. London: Duxbury Press, 1996.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Sipe, M.A., Casasent, D. Global feature space neural network for active computer vision. Neural Comput & Applic 7, 195–215 (1998). https://doi.org/10.1007/BF01414882
Issue Date:
DOI: https://doi.org/10.1007/BF01414882