Journal of the Korean Institute of Intelligent Systems (한국지능시스템학회논문지)

Korean Institute of Intelligent Systems (한국지능시스템학회)
권호 표지
DOI QR코드

DOI QR Code

Separations and Feature Extractions for Image Signals Using Independent Component Analysis Based on Neural Networks of Efficient Learning Rule

효율적인 학습규칙의 신경망 기반 독립성분분석을 이용한 영상신호의 분리 및 특징추출

  • 조용현 (대구가톨릭대학교 컴퓨터정보통신공학부)
  • Published : 2003.04.01
Download PDF
⟨ Previous Next ⟩

Abstract

This paper proposes a separation and feature extraction of image signals using the independent component analysis(ICA) based on neural networks of efficient learning rule. The proposed learning rule is a hybrid fixed-point(FP) algorithm based on secant method and momentum. Secant method is applied to improve the performance by simplifying the 1st-order derivative computation for optimizing the objective function, which is to minimize the mutual informations of the independent components. The momentum is applied for high-speed convergence by restraining the oscillation in the process of converging to the optimal solution. The proposed algorithm has been applied to the composite images generated by random mixing matrix from the 10 images of $512\times512$-pixel. The simulation results show that the proposed algorithm has better performances of the separation speed and rate than those using the FP algorithm based on Newton and secant method. The proposed algorithm has been also applied to extract the features using a 3 set of 10,000 image patches from the 10 fingerprints of $256\times256$-pixel and the front and the rear paper money of $480\times225$-pixel, respectively, The simulation results show that the proposed algorithm has also better extraction speed than those using the another methods. Especially, the 160 basis vectors(features) of $16\times16$-pixel show the local features which have the characteristics of spatial frequency and oriented edges in the images.

본 연구에서는 효율적인 학습규칙의 신경망 기반 독립성분분석기법을 이용한 영상신호의 분리와 특징추출을 제안하였다. 제안된 학습규칙은 할선법과 모멘트를 이용한 조합형 고정점 학습알고리즘이다. 여기서 할선법은 독립성분 상호간의 정보를 최소화하기 위한 목적함수의 최적화 과정에서 요구되는 1차 미분에 따른 계산을 간략화하기 위함이고, 모멘트는 최적화 과정에서 발생하는 발진을 억제하여 보다 빠른 학습을 위함이다. 제안된 기법을 $512\times512$의 픽셀을 가지는 10개의 영상을 대상으로 임의의 혼합행렬에 따라 발생되는 혼합영상의 분리에 적용한 결과, 뉴우턴법에 기초한 기존의 알고리즘과 할선법만에 기초한 알고리즘보다 각각 우수한 분리률과 빠른 분리속도가 있음을 확인하였다. 또한 $256\times256$ 픽셀의 10개 지문상과 $480\times225$ 픽셀의 지폐영상에서 선택된 각각 10,000개의 3가지 영상패치들을 대상으로 적용한 결과, 제안된 기법은 뉴우턴법이나 할선법의 알고리즘보다도 빠른 특징추출 속도가 있음을 확인하였다. 한편 추출된 $16\times16$ 펙셀의 160개 독립성분 기저벡터 각각은 영상 각각에 포함된 공간적인 주파수 특성과 방향성을 가지는 경계 특성이 잘 드러나는 국부적인 특징들임을 확인하였다.

Keywords

References

  1. K. I. Diamantara sand S. Y. Kung, Principal Component Neural Networks : Theory and Applications, Adaptive and learning Systems for Signal Processing, Communications, and Control, John Wiley & Sons, Inc., 1996
  2. S. Haykin, Neural Networks : A Comprehensive Foundation, Prentice-Hall, 2ed, London, 1999
  3. J. Karhunen and J. Joutsensalo, "Generation of Principal Component Analysis, Optimization Problems, and Neural Networks," Neural Networks, vol. 8, No. 4, pp. 549-562, 1995 https://doi.org/10.1016/0893-6080(94)00098-7
  4. P. Comon, "Independent Component Analysis A New Concept?", Signal Processing, vol. 36, No. 3, pp. 287-314, Apr. 1994 https://doi.org/10.1016/0165-1684(94)90029-9
  5. T. W. Lee, Independent Component Analysis : Theory and Applications, Kluwer Academic Pub., Boston, 1998
  6. J. Karhunen, "Neural Approaches to Independent Component Analysis and Source Seperation", 4th European Symp., Artificial Neural Network, ESANN96, Burges, Belgium, pp. 249-266, Apr. 1996
  7. S. Roberts and R. Everson, Independent Component Analysis : Principles and Practice, Cambridge Univ. Press, 2001
  8. A. Hyvarinen, J. Karhunen, and E. Oja, Independent Component Analysis, John Wiley & Sons, Inc., 2001
  9. A. J. Bell and T. J. Sejnowski, "Edges are 'Independent Components' of Natural Scenes", "Advances in Neural Information Processing Systems 9", MIT Press, 1996
  10. K. Atkinson, Elementary Numerical Analysis', John Wiley & Sons, Inc., New York, 1993