Journal of KIISE:Software and Applications (한국정보과학회논문지:소프트웨어및응용)

Korean Institute of Information Scientists and Engineers (한국정보과학회)
권호 표지

2D Pattern Deformation Analysis using Particle and Spring-Damper Mesh

입자와 스프링-댐퍼 메쉬를 이용한 2차원 패턴 변형 분석

  • 신봉기 (부경대학교 전자컴퓨터정보통신공학부)
  • Published : 2005.08.01
Download PDF
⟨ Previous Next ⟩

Abstract

This paper addresses a novel application of meshes to analyzing the deformation patterns of 2D signals. The proposed mesh is distinguished form the previous models in that it includes simulated charges in each node that interact with external charges comprising an input pattern. Therelaxation of the mesh given an input is carried out by any of the well-known numerical integration techniques. The result of the relaxation is a deformed mesh. This Paper provides four criterion functions for measuring the pattern deformation. A set of trained meshes was created from the simple average of target patterns. Experimental results show that these measures, although highly intuitive, are not good enough to capture the amount and characteristics of pattern deformation. If more sophisticated measures are found and incorporated into the relaxation process, we expect that a better and high-performance mesh framework is realized.

본 논문에서는 스프링-댐퍼 메쉬를 2차원 영상 패턴의 변형을 모델링 및 분석하는 데 응용하는 방법을 제안한다. 기존의 메쉬와 다르게 새롭게 제안된 메쉬 모형은 메쉬 노드와 영상 픽셀을 하전입자로 표현하고 둘 사이의 상호작용에 의하여 변형되는 메쉬를 다양한 기준함수에 의하여 분석하는 것이다. 메쉬의 변형 과정은 하전입자 사이의 상호작용에 의한 스트레스 완화 과정으로써 널리 알려진 간단한 수치적분법을 사용한다. 변형 메쉬평가함수로 네 가지를 제안하고 각각의 성능을 분석하였다. 훈련 메쉬로는 샘플 영상의 평균 패턴을 구하여 메쉬로 직접 변환한 것을 사용하였다. 실험 결과 각 함수는 국소 변형을 모두 제대로 평가하는 측도로서는 아직 부족하지만 앞으로 이에 대한 보완 또는 새로운 함수의 제안, 그리고 이완 과정의 개선이 뒤따른다면 보다 체계적인 분석 방법과 높은 성능을 제공할 것으로 판단된다.

Keywords

References

  1. B. K. P. Horn, B. B. Schunck, 'Determining optical flow,' Artificial Intelligence, Vol. 17, pp. 185-203, 1981 https://doi.org/10.1016/0004-3702(81)90024-2
  2. M. Kass, A. Witkin and D. Terzopoulos 'Snakes, active contour models,' International Journal of Computer Vision, Vol.1, 1987, pp.321-331 https://doi.org/10.1007/BF00133570
  3. S. -Y. Kim and S. -W. Lee, 'Gray-Scale Nonlinear Shape Normalization Method for Handwritten Oriental Character Recognition,' Int. J. PRAI, vol. 12, no. 1, 1998, pp.81-95 https://doi.org/10.1142/S0218001498000075
  4. D.-H. Lee and H.-G. Cho, 'A new synthesizing method for handwriting script,' Int. J. of Pattern Recognition and Artificial Intelligence, vol. 12, no. 1, pp.45-61, 1998 https://doi.org/10.1142/S0218001498000051
  5. A. Baraff and D. Witkin, 'Large steps in cloth simulation,' in Proc. SIGGRAPH, pp.43-54, 1998
  6. T. Wakahara and K. Odaka, 'Adaptive normalization of handwritten characters using global/local affine transformation,' IEEE Trans. PAMI, vol. 20, no. 12, Dec. 1998 https://doi.org/10.1109/34.735806
  7. M. Revow, C. Williams, and G. Hinton, 'Using generative models for handwritten digit recognition,' IEEE Trans. PAMI, vol. 18, no. 6, pp. 592-606, June 1996 https://doi.org/10.1109/34.506410
  8. K.-W. Cheung, D.-Y. Yeung, and R. Chin, 'A Bayesian framework for deformable pattern recognition with application to handwritten character recognition,' IEEE Trans. PAMI, vol. 20, no. 12, Dec. 1998 https://doi.org/10.1109/34.735813
  9. A. Jain and D. Zongker, 'Representation and recognition of handwritten digits using deformable templates,' IEEE Trans. PAMI, vol. 19, no. 12, pp. 1386-1390, Dec. 1997 https://doi.org/10.1109/34.643899
  10. R. Webster, T. Nagasaki, T. Teramura, and M. Nakagawa, 'Several possibilities for character recognition based on a dynamic model,' in The 4th IWFHR, Taipei, pp. 423-430, Dec. 1994
  11. D. Doermann and A. Rosenfeld, 'Recovery of temporal information from static images of handwriting,' Computer Vision and Pattern Recognition, 1992 https://doi.org/10.1109/CVPR.1992.223211
  12. Geman and D.Geman, 'Stochastic relaxation, Gibbs distributions and the Bayesian restoration of images,' IEEE Trans. PAMI, 6:721-741, 1984