This Article 
   
 Share 
   
 Bibliographic References 
   
 Add to: 
 
Digg
Furl
Spurl
Blink
Simpy
Google
Del.icio.us
Y!MyWeb
 
 Search 
   
Cooperative Vision Integration Through Data-Parallel Neural Computations
December 1991 (vol. 40 no. 12)
pp. 1368-1379

The authors describe a neural network approach for combining processing of multiple early vision modules. Energy functions for coupling the computation of intensity contours, optical flow, and stereo disparity are defined. Hopfield neural networks are used for function minimization with deterministic annealing to avoid spurious local minima. Vision integration schemes are developed by extending the work of T.A. Poggio et al. (1988) to include cooperative interactions between different vision modules and the Hebbian adaptation of vision module coupling on a massively parallel computer consisting of 4096 processing elements operated in a single-instruction-multiple-data mode. Simple experiments assess the performance of various integration approaches. The resulting algorithms facilitate fast, robust image segmentation.

[1] J. Aloimonos and D. Shulman,Integration of Visual Modules: An Extension of the Marr Paradigm. Boston, MA: Academic, 1989.
[2] S. T. Barnard, "Stochastic stereo matching over scale,"Int. J. Comput. Vision, vol. 3, pp. 17-32, 1989.
[3] H. B. Barrow and J. M. Tenenbaum, "Recovering intrinsic scene characteristics from images," inComputer Vision Systems, A. R. Hanson and E. M. Riseman, Eds. New York: Academic, 1978.
[4] M. Bertero, T. A. Poggio, and V. Torre, "Ill-posed problems in early vision,"Proc. IEEE, vol. 76, pp. 869-889, Aug. 1988.
[5] D. P. Bertsekas and J. N. Tsitsiklis,Parallel and Distributed Computations. Englewood Cliffs, NJ: Prentice-Hall, 1989.
[6] A. Blake and A. Zisserman,Visual Reconstruction. Cambridge, MA: MIT Press, 1987.
[7] J. J. Clark and A. L. Yuille,Data Fusion for Sensory Information Processing Systems. Boston, MA: Kluwer, 1990.
[8] M. J. Daily, "Self-calibration of a multi-camera vision system," inProc. 24th Asilomar Conf. Signals, Syst., Comput., Monterey, CA, Nov. 1990.
[9] P. M. Dew, R. A. Earnshaw, and T. R. Heywood, Eds.,Parallel Processing for Computer Vision and Display. Wokingham, England: Addison-Wesley, 1989.
[10] K. Fukushima, "A neural network model for selective attention in visual pattern recognition,"Biol. Cybern., vol. 55, pp. 5-15, 1986.
[11] E. B. Gamble, D. Geiger, T. A. Poggio, and D. Weinshall, "Integration of vision modules and labeling of surface discontinuities,"IEEE Trans. Syst., Man, Cybern., vol. 19, pp. 1575-1581, 1989.
[12] E. B. Gamble and T. Poggio, "Visual integration and detection of discontinuities: The key role of intensity edges," Artificial Intell. Lab., Massachusetts Inst. Technol., AI Memo 970, Oct. 1987.
[13] D. Geiger and F. Girosi, "Parallel and deterministic algorithms from mrf's: surface reconstruction,"IEEE Trans. Pattern Anal. Machine Intell., vol. 13, pp. 401-412, May 1991.
[14] S. Geman and D. Geman, "Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images,"IEEE Trans. Pattern Anal. Machine Intell., vol. PAMI-6, pp. 721-741, June 1984.
[15] W. E. L. Grimson, "Binocular shading and visual surface reconstruction,"Computer Vision, Graphics, Image Processing, pp. 18-44, 1984.
[16] L. A. Hageman and D. M. Young,Applied Iterative Methods. New York: Academic, 1981.
[17] J. Harris, C. Koch, and J. Luo, "A two-dimensional analog vlsi circuit for detecting discontinuities in early vision,"Science, pp. 1209-1211, 1990.
[18] J. Hertz, A. Krogh, and R. G. Palmer.Introduction to the Theory of Neural Networks. Reading, MA: Addison-Wesley, 1991.
[19] J. J. Hopfieldand D. W. Tank, "Neural computation of decision in optimization problems,"Biol. Cybern., vol. 52, pp. 141-152, 1985.
[20] B. K. P. Horn,Robot Vision. Cambridge, MA: M.I.T. Press, 1986.
[21] B. K. P. Horn and B. G. Schunck, "Determining optic flow,"Artif. Intell., vol. 17, pp. 185-203, 1981.
[22] J. Hutchinson, C. Koch, J. Juo, and C. Mead, "Computing motion using analog and resistive networks,"IEEE Comput. Mag., pp. 52-63, Mar. 1988.
[23] C. Koch, J. Marroquin, and A. Yuille, "Analog `neuronal' networks in early vision," AI Memo 751, M.I.T. Artificial Intelligence Lab., June 1985.
[24] D. G. Luenberger,Linear and Nonlinear Programming. Reading, MA: Addison-Wesley, 1986.
[25] R. C. Luo and M. G. Kay, "Multisensor integration and fusion in intelligent systems,"IEEE Trans. Syst., Man, Cybern., vol. 19, pp. 901-931, Sept./Oct. 1989.
[26] D. Marr,Vision. San Francisco, CA: Freeman, 1982.
[27] D. W. Murray, A. Kashko, and H. Buxton, "A parallel approach to the picture restoration algorithm of geman and geman on an simd machine."Image and Vision Comput., vol. 4, no. 3, pp. 133-142, Aug. 1986.
[28] N. M. Nasrabadi, S. P. Clifford, and Y. Liu, "Integration of stereo vision and optical flow by using an energy minimization approach,"J. Opt. Soc. Amer. A, vol. 6, no. 6, pp. 900-907, June 1989.
[29] D. Parkinson and J. Litt, Eds.,Massively Parallel Computing with the DAP. Cambridge, MA: M.I.T Press, 1990.
[30] T. A. Poggio, E. B. Gamble, and J. J. Little, "Parallel integration of vision modules,"Science, vol. 242, pp. 436-440, Oct. 1988.
[31] T. Poggio and F. Girosi, "A theory for approximation and learning," AI Memo 1140, MIT, July 1989.
[32] T. A. Poggio and C. Koch, "Ill-posed problems in early vison: From computational theory to analogue networks,"Proc. Royal Soc. London B, vol. 266, pp. 303-323, 1985.
[33] A. Rangarajan and R. Chellappa, "Generalized graduated nonconvexity algorithm for maximum a posteriori image estimation," Tech. Rep. 149, U.S.C. Signal and Image Processing Institute, Dec. 1989.
[34] D. Terzopoulos, "Integrating visual information from multiple sources," inFrom Pixels to Predicates, C. M. Brown, Ed. Norwood, NJ: Ablex, 1989, pp. 111-142.
[35] S. T. Toborg, "A 3-d wafer scale architecture for early vision processing," inProc. Int. Conf. Appl. Specific Array Processors, Princeton, NJ, IEEE Computer Society Press, 1990, pp. 246-258.
[36] S. T. Toborg and K. Hwang, "Exploring neural network and optical computing technologies," inParallel Processsing for Supercomputing and Artificial Intelligence, Hwang and DeGroot, Eds. New York: McGraw-Hill, 1989, ch. 16.
[37] J. N. Tsitsiklis, "A comparison of Jacobi and Gauss-Seidel parallel iterations," M.I.T. Industrial Liaison Program Rep. LIDS-P-1837, M.I.T. Labo. for Information and Decision Systems, Nov. 1988.
[38] A. L. Yuille, "Energy functions for early vision and analog networks,"Biol. Cybern., vol. 61, pp. 115-123, 1989.

Index Terms:
cooperative vision integration; Hopfield neural networks; data-parallel neural computations; neural network approach; multiple early vision modules; intensity contours; optical flow; stereo disparity; function minimization; deterministic annealing; Hebbian adaptation; massively parallel computer; single-instruction-multiple-data mode; image segmentation; computer vision; computerised picture processing; neural nets.
Citation:
S.T. Toborg, K. Hwang, "Cooperative Vision Integration Through Data-Parallel Neural Computations," IEEE Transactions on Computers, vol. 40, no. 12, pp. 1368-1379, Dec. 1991, doi:10.1109/12.106222
Usage of this product signifies your acceptance of the Terms of Use.