Abstract

Pattern-recognition problems for which patterns cannot be recognized directly but by their attitudes and/or behaviors is addressed. To analyze these attitudes, pattern signatures are generated from picture sequences. Two complementary signature synthesis algorithms are presented. The architecture is made up of two cascaded correlators. The first is used to create the signatures and the second to classify them. We focus our analysis on the case of optical implementations. Illustrations are given in the case of face recognition by attitudes (multisensor in the optronic imaging range) and moving-target recognition by behavior (in the radar imaging range).

© 2000 Optical Society of America

Full Article  |  PDF Article

References

  • View by:
  • |
  • |
  • |

  1. G. Keryer, J.-L. de Bougrenet de la Tocnaye, A. Al Falou, “Performance comparison of ferroelectric liquid-crystal-technology based multichannel correlators,” Appl. Opt. 36, 3043–3055 (1997).
    [CrossRef] [PubMed]
  2. C. Elachi, T. Bicknell, R. L. Jordan, Ch. Wu, “Spaceborne synthetic-aperture imaging radars: applications, techniques, and technology,” Proc. IEEE 70, 1174–1209 (1982).
    [CrossRef]
  3. B. V. K. Vijaya Kumar, I. Hassebrook, “Performance measures for correlation filters,” Appl. Opt. 29, 2997–3006 (1990).
    [CrossRef]
  4. B. Javidi, “Nonlinear power spectrum based optical correlation,” Appl. Opt. 28, 2358–2367 (1989).
    [CrossRef] [PubMed]
  5. H.-O. Peitgen, H. Jürgens, D. Saupe, Chaos and Fractals (Springer-Verlag, New York, 1992).
    [CrossRef]
  6. Ph. Réfrégier, V. Laude, B. Javidi, “Nonlinear joint transform correlator: an optimal solution for adaptive image discrimination and input noise robustness,” Opt. Lett. 10, 405–407 (1994).
  7. M. Sakalli, H. Yan, “Feature based compression human face images,” Opt. Eng. 37, 1520–1529 (1998).
    [CrossRef]
  8. T.-H. Chao, B. Lau, A. Yacoubian, “Mine detection using wavelet processing of electro-optic active sensor data,” in Detection Technologies for Mines and Minelike Targets, A. C. Dubey, I. Cindrich, J. M. Ralston, K. A. Rigano, eds., Proc. SPIE2496, 433–441 (1995).
    [CrossRef]
  9. J.-L. de Bougrenet de la Tocnaye, E. Quémener, Y. Pétillot, “Composite versus multichannel binary phase-only filtering,” Appl. Opt. 36, 6646–6653 (1997).
    [CrossRef]
  10. M. Killinger, J.-L. de Bougrenet de la Tocnaye, W. A. Crossland, “Bistability and nonlinearity in optically addressed ferroelectric liquid-crystal spatial light modulators: applications to neurocomputing,” Appl. Opt. 31, 3930–3936 (1992).
    [CrossRef] [PubMed]
  11. P. Miller, “Multiresolution correlator analysis and filter design,” Appl. Opt. 35, 5790–5810 (1996).
    [CrossRef] [PubMed]
  12. H. Szu, Y. Sheng, J. Chen, “Wavelet transform as a bank of the matched filters,” Appl. Opt. 31, 3267–3277 (1992).
    [CrossRef] [PubMed]

1998 (1)

M. Sakalli, H. Yan, “Feature based compression human face images,” Opt. Eng. 37, 1520–1529 (1998).
[CrossRef]

1997 (2)

1996 (1)

1994 (1)

1992 (2)

1990 (1)

1989 (1)

1982 (1)

C. Elachi, T. Bicknell, R. L. Jordan, Ch. Wu, “Spaceborne synthetic-aperture imaging radars: applications, techniques, and technology,” Proc. IEEE 70, 1174–1209 (1982).
[CrossRef]

Al Falou, A.

Bicknell, T.

C. Elachi, T. Bicknell, R. L. Jordan, Ch. Wu, “Spaceborne synthetic-aperture imaging radars: applications, techniques, and technology,” Proc. IEEE 70, 1174–1209 (1982).
[CrossRef]

Chao, T.-H.

T.-H. Chao, B. Lau, A. Yacoubian, “Mine detection using wavelet processing of electro-optic active sensor data,” in Detection Technologies for Mines and Minelike Targets, A. C. Dubey, I. Cindrich, J. M. Ralston, K. A. Rigano, eds., Proc. SPIE2496, 433–441 (1995).
[CrossRef]

Chen, J.

Crossland, W. A.

de Bougrenet de la Tocnaye, J.-L.

Elachi, C.

C. Elachi, T. Bicknell, R. L. Jordan, Ch. Wu, “Spaceborne synthetic-aperture imaging radars: applications, techniques, and technology,” Proc. IEEE 70, 1174–1209 (1982).
[CrossRef]

Hassebrook, I.

Javidi, B.

Jordan, R. L.

C. Elachi, T. Bicknell, R. L. Jordan, Ch. Wu, “Spaceborne synthetic-aperture imaging radars: applications, techniques, and technology,” Proc. IEEE 70, 1174–1209 (1982).
[CrossRef]

Jürgens, H.

H.-O. Peitgen, H. Jürgens, D. Saupe, Chaos and Fractals (Springer-Verlag, New York, 1992).
[CrossRef]

Keryer, G.

Killinger, M.

Lau, B.

T.-H. Chao, B. Lau, A. Yacoubian, “Mine detection using wavelet processing of electro-optic active sensor data,” in Detection Technologies for Mines and Minelike Targets, A. C. Dubey, I. Cindrich, J. M. Ralston, K. A. Rigano, eds., Proc. SPIE2496, 433–441 (1995).
[CrossRef]

Laude, V.

Miller, P.

Peitgen, H.-O.

H.-O. Peitgen, H. Jürgens, D. Saupe, Chaos and Fractals (Springer-Verlag, New York, 1992).
[CrossRef]

Pétillot, Y.

Quémener, E.

Réfrégier, Ph.

Sakalli, M.

M. Sakalli, H. Yan, “Feature based compression human face images,” Opt. Eng. 37, 1520–1529 (1998).
[CrossRef]

Saupe, D.

H.-O. Peitgen, H. Jürgens, D. Saupe, Chaos and Fractals (Springer-Verlag, New York, 1992).
[CrossRef]

Sheng, Y.

Szu, H.

Vijaya Kumar, B. V. K.

Wu, Ch.

C. Elachi, T. Bicknell, R. L. Jordan, Ch. Wu, “Spaceborne synthetic-aperture imaging radars: applications, techniques, and technology,” Proc. IEEE 70, 1174–1209 (1982).
[CrossRef]

Yacoubian, A.

T.-H. Chao, B. Lau, A. Yacoubian, “Mine detection using wavelet processing of electro-optic active sensor data,” in Detection Technologies for Mines and Minelike Targets, A. C. Dubey, I. Cindrich, J. M. Ralston, K. A. Rigano, eds., Proc. SPIE2496, 433–441 (1995).
[CrossRef]

Yan, H.

M. Sakalli, H. Yan, “Feature based compression human face images,” Opt. Eng. 37, 1520–1529 (1998).
[CrossRef]

Appl. Opt. (7)

Opt. Eng. (1)

M. Sakalli, H. Yan, “Feature based compression human face images,” Opt. Eng. 37, 1520–1529 (1998).
[CrossRef]

Opt. Lett. (1)

Proc. IEEE (1)

C. Elachi, T. Bicknell, R. L. Jordan, Ch. Wu, “Spaceborne synthetic-aperture imaging radars: applications, techniques, and technology,” Proc. IEEE 70, 1174–1209 (1982).
[CrossRef]

Other (2)

H.-O. Peitgen, H. Jürgens, D. Saupe, Chaos and Fractals (Springer-Verlag, New York, 1992).
[CrossRef]

T.-H. Chao, B. Lau, A. Yacoubian, “Mine detection using wavelet processing of electro-optic active sensor data,” in Detection Technologies for Mines and Minelike Targets, A. C. Dubey, I. Cindrich, J. M. Ralston, K. A. Rigano, eds., Proc. SPIE2496, 433–441 (1995).
[CrossRef]

Cited By

OSA participates in CrossRef's Cited-By Linking service. Citing articles from OSA journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (12)

Fig. 1
Fig. 1

Generic architecture, characterized by two blocks and three layers: observation and measurement layer, filter and signature basis layer, and processing layer. The first block creates the signatures, and the second one recognizes this signature.

Fig. 2
Fig. 2

Simplified architectures: MS and PCE (MS–PCE), MS and SF curve (MS–SF), and MR and SF curve (MR–SF).

Fig. 3
Fig. 3

Hc used for scanning the cross-correlation plane: a Hilbert path of 256 pixels fills a 16 × 16 pixel image.

Fig. 4
Fig. 4

Examples of attitude for Emmanuel and bearded bases, including nod as the learning attitude and nod and yawn as the test attitude.

Fig. 5
Fig. 5

Examples of images and their corresponding filters for two different faces (Emmanuel and bearded). BPOF, binary POF.

Fig. 6
Fig. 6

Signatures obtained by the MS–PCE approach. Left-hand column, signatures obtained for a nod attitude, and right-hand column, signatures obtained for a nod and yawn attitude, for two different faces (E, Emmanuel; B, Bearded).

Fig. 7
Fig. 7

On the top appears the cross correlations between visible and infrared images for each sequence element. We extracted a curve from each cross-correlation plane by following a SF curve (Hc) that forms a signature line.

Fig. 8
Fig. 8

Signatures obtained by a MS–SF approach. Left-hand column, signatures obtained for a nod, and right-hand column, signatures obtained for a nod and yawn, for two different faces (E, Emmanuel; and B, bearded). For a signature, each line holds the result of the cross correlation between infrared and visible images of input sequence elements.

Fig. 9
Fig. 9

Examples of RGD radar images obtained for three different boats: tugboat (TB), trawler (TR), and oil tanker (OT). Two successive images of the learning and the test bases are presented. The deviation between two successive images can be noted for each boat.

Fig. 10
Fig. 10

Stochastic attitude for TB, TR, and OT bases.

Fig. 11
Fig. 11

Steps in MR–SF architecture: FT’s are used to change space. In spectral space the result is multiplied by a filter and raised to a power elevation. After these operations, in direct space, the cross-correlation part is extracted with a SF curve. Curve, path following Hc.

Fig. 12
Fig. 12

Signatures obtained by a MR–SF approach. Left-hand column, signatures obtained for the first 18 RGD images, and right-hand column, signatures obtained for the final 18 images, for three different boats: TB, TR, and OT. For a signature, each line holds the result from the cross correlation between the first image with each element of the input sequence.

Tables (3)

Tables Icon

Table 1 Nod and Yawn Signatures Obtained by MS–PCE Approach for Two Faces (Emmanuel and Bearded) and Compared by Correlation with the Bp Signature Basis, for Each Wavelet Order la

Tables Icon

Table 2 Nod and Yawn Signatures Obtained by MS–SF Approach for Two Faces (Emmanuel and Bearded) and Compared by Correlation with the Bp Signature Basis, for Each Wavelet Order la

Tables Icon

Table 3 Signatures Obtained by a MR–SF Approach for Three Boats (TB, TR, and OT) and Compared by Correlation with the Bp Signature Basis, for Each Wavelet Order la

Equations (12)

Equations on this page are rendered with MathJax. Learn more.

Fm=Xm*|Xm|2k,
Fml=Fm*|Gl||Fm|.
FT-1cml=Xi, YiFml
Ciml=FT|Xi, YiFml|2k.
Silr=SFFT|Xi, Yi|2|Gl|2.
PCEp=l σlPCEplγl.
ci+j,ml=|Xi, Xi+j×Fml|2k.
ci+j,l=|Xi, Xi+j|×|Gl|2k.
cml=xglfmgl,
Cml=X×Gl*×Fm×Gl**=X×Fm*|Gl|2
=X×Fml with  Fml=Fm*×|Gl|2.
ψax, y=1|a|43π1-x2a21-y2a2exp-x2+y22a2.

Metrics