Abstract
The dynamic-wire methodology provides dedicated lines of communication among groups of pixels of an image which share common properties. In simple applications, object regions can be grouped together to compute the area or the center of mass of each object. Alternatively, object boundaries may be used to compute curvature or contour length. These measurements are useful for higher-level tasks such as object recognition or structural saliency. The dynamic-wire methodology is efficiently implemented in fast, low-power analog hardware. Switches create a true electrical connection among selected pixels, dynamically configuring wires or resistive networks on the fly. Dynamic wires provide a model for object-based processing. This approach is different from present early vision chips which are limited to pixel-based or image-based operations. Using this methodology, we have successfully designed and demonstrated a custom analog VLSI chip which computes contour length.
Similar content being viewed by others
References
H., Asada and M., Brady, The curvature primal sketch, IEEE Trans. Patt. Anal. Mach. Intell. 8(1): 2–14, 1986.
W., Bair and C., Koch, An analog VLSI chip for finding edges from zero-crossing. In D., Touretzky, ed., Neural Information Processing Systems. Morgan-Kaufmann: Palo Alto, 1991.
A.R., Bruss and B.K.P., Horn, Passive navigation, Comput. Vis. Graph. Image Process. 21(1): 3–20, January 1983.
S.P., DeWeerth and C.A., Mead, A two-dimensional visual tracking array. In 1988 MIT Conference on Very Large Scale Integration, MIT Press: Cambridge, MA, pp. 259–275, 1988.
J.M. Hakkarainen and J.J. Little, Interaction of algorithm and implementation for analog VLSI stereo. In SPIE Technical Symposia on Optical Engineering and Photonics in Aerospace Sensing, Orlando, FL, April 1991. Visual Information Processing: From Neurons to Chips.
J.G., Harris, C., Koch, and J., Luo, A two-dimensional analog VLSI circuit for detecting discontinuities in early vision, Science 248: 1209–1211, 1990.
J.G., Harris, C., Koch, J., Luo, and J., Wyatt, Resistive fuses: analog hardware for detecting discontinuities in early vision. In C., Mead and M., Ismail, eds., Analog VLSI Implementations of Neural Systems, Kluwer: Norwell, MA, 1989.
B.K.P., Horn, Robot Vision. MIT Press: Cambridge, MA, 1986.
B.K.P., Horn and E.J., Weldon, Filtering closed curves, IEEE Trans. Patt. Anal. Mach. Intell. 8(5): 665–668, 1986.
T. Knight, Design of an integrated optical sensor with on-chip preprocessing. Ph.D. thesis, MIT, Cambridge, MA, 1983. Dept. of Electrical Engineering and Computer Science.
H., Kobayashi, J.L., White, and A.A., Abidi, An active resistor network for Gaussian filtering of images. IEEE J. Solid-State Circuits 26(5): 738–748, May 1991.
J., Lazzaro, S., Ryckebusch, M.A., Mahowald, and C.A., Mead, Winner-take-all networks of o(n) complexity. In D., Touretzky, ed., Advance in Neural Information Processing Systems, vol. 1, Morgan Kaufmann: Palo Alto, CA, pp. 703–711, 1989.
D. Lowe, Organization of smooth image curves at multiple scales. Proc. Conf. Comput. Vis. Patt. Recogn., Ann Arbor, 1988.
J. Luo, C. Koch, and C. Mead, An experimental subthreshold, analog CMOS two-dimensional surface interpolation circuit. Neural Information Processing Systems Conference, Denver, CO, November 1989.
M. Mahowald and T. Delbrück, An analog VLSI implementation of the marr-poggio stereo correspondence algorith, Proc. Intell. Neural Network Soc., Boston, MA, 1988.
I.S. McQuirk, Direct methods for estimating the focus of expansion in analog VLSI. Master's thesis, Massachusetts Institute of Technology, Cambridge, MA, September 1991. Department of Electrical Engineering and Computer Science.
C., Mead, Analog VLSI and Neural Systems. Addison-Wesley: Reading, MA, 1989.
C., Mead and T., Delbrück, Scanners for visualizing activity of analog VLSI circuitry. In Analog VLSI and Signal Processing. Kluwer Academic: Norwell, MA, 1991.
C.A., Mead and M.A., Mahowald, A silicon model of early visual processing, Neural Networks 1: 91–97, 1988.
F., Mokhtarian and A.K., Mackworth, Scale-based description and recognition of planar curves and two-dimensional shapes, Trans. Patt. Anal. Mach. INtell. 8(1): 34–43, 1986.
J.P., Sage and A.L., Lattes, A high-speed two-dimensional CCD gaussian image convolver. Quart. Tech. Rept., MIT Lincoln Laboratory, Lexington, MA, 1987, August–October, 1986.
A. Shaashua and S. Ullman, Structural saliency: The detection of globally salient structures using a locally connected network, Proc. Comput. Vis. Patt. Recog. Conf., Ann Arbor, 1988.
D.B., Shu, J.G., Nash, M.M., Eshaghain, and K., Kim, Implementation and application of a gated-connection network in image understanding. In H., Li and Q.F., Stout, eds., Reconfigurable Massively Parallel Computers, Prentice-Hall: Englewood Cliffs, NJ, pp. 64–87, 1991.
D. Standley, Analog VLSI implementation of smart vision sensors: stability theory and an experimental design, Ph.D. thesis, MIT, Cambridge, MA, 1991. Dept. of Electrical Engineering and Computer Science.
J.E., Tanner and C., Mead. An integrated analog optical motion sensor. In S.Y., Kung, ed., VLSI Signal Processing, II, IEEE Press: New York, p. 59–76, 1986.
W. Yang, A charge-coupled device architecture for on focal plane image signal processing, Proc. Intern. Symp. VLSI Tech., Syst., Appl., pp. 266–270, May 1990.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Liu, SC., Harris, J. Dynamic wires: An alanog VLSI model for object-based processing. Int J Comput Vision 8, 231–239 (1992). https://doi.org/10.1007/BF00055154
Issue Date:
DOI: https://doi.org/10.1007/BF00055154