This Article 
 Bibliographic References 
 Add to: 
Learning an Integral Equation Approximation to Nonlinear Anisotropic Diffusion in Image Processing
April 1997 (vol. 19 no. 4)
pp. 342-352

Abstract—Multiscale image enhancement and representation is an important part of biological and machine early vision systems. The process of constructing this representation must be both rapid and insensitive to noise, while retaining image structure at all scales. This is a complex task as small scale structure is difficult to distinguish from noise, while larger scale structure requires more computational effort. In both cases, good localization can be problematic. Errors can also arise when conflicting results at different scales require cross-scale arbitration.

Broadly speaking, multiscale image analysis has historically been accomplished using two types of techniques: those which are sensitive to image structure and those which are not. Algorithms in the latter category typically use a set of variously sized blurring kernels to produce images, each of which retains structure at a different scale [1], [2], [3], [4]. The kernels used for the blurring are predefined and independent of the content of the image. Koenderink showed that if the kernels are Gaussian, then this process is equivalent to the evolution of the linear heat (or diffusion) equation. He thus transformed the integral equation representing the convolution process into the solution of a partial differential equation (PDE).

Structure sensitive multiscale techniques attempt to analyze an image at a variety of scales within a single image [5], [6], [7]. Klinger [5] proposed the quad tree, one of the earliest structure-sensitive multiscale image representations. In this approach, a tree structure is built by recursively subdividing an image based on pixel variance in subregions. The final tree contains leaves representing image regions whose variance is small according to some measure. Recently [6], [8], the PDE formalism introduced by Koenderink has been extended to allow structure-sensitive multiscale analysis. Instead of the uniform blurring of the linear heat equation which destroys small scale structure as time evolves, Perona and Malik use a space-variant conductance coefficient based on the magnitude of the intensity gradient in the image, giving rise to a nonlinear PDE. Like the quadtree, the end result is a single image representation which contains information at all scales of interest.

The Perona and Malik approach produces impressive results, but the numerical integration of a nonlinear PDE is a costly and inherently serial process. In this paper, we present a technique which obtains an approximate solution to the PDE for a specific time, via the solution of an integral equation which is the nonlinear analog of convolution. The kernel function of the integral equation plays the same role that a Green's function does for a linear PDE, allowing the direct solution of the nonlinear PDE for a specific time without requiring integration through intermediate times. We then use a learning technique to approximate the kernel function for arbitrary input images. The result is an improvement in speed and noise-sensitivity, as well as providing a means to parallelize an otherwise serial algorithm.

[1] D. Marr and E. Hildreth, "Theory of Edge Detection," Proc. Royal Soc. of London, vol. B207, pp. 187-217, 1980.
[2] P.J. Burt and E.H. Adelson, “The Laplacian Pyramid as a Compact Image Code,” IEEE Trans. Comm., vol. 31, no. 4, pp. 532-540, 1983.
[3] J. Koenderink, "The Structure of Images," Biological Cybernetics, vol. 50, pp. 363-370, 1984.
[4] R.A. Hummel, "Representations Based on Zero-Crossings in Scale-Space," Readings in Computer Vision: Issues, Problems, Principles and Paradigms, M. Fischler and O. Firschein, eds. Los Angeles: Morgan Kaufmann, 1986.
[5] A. Klinger, "Pattern and Search Statistics," Optimizing Methods in Statistics.New York: Academic Press, 1971.
[6] P. Perona and J. Malik, "Scale Space and Edge Detection Using Anisotropic Diffusion," Proc. IEEE Workshop Computer Vision,Miami, Fla., 1987, pp. 16-27.
[7] M. Nitzberg and T. Shiota, “Nonlinear Image Filtering with Edge and Corner Enhancement,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 14, pp. 826-833, 1992.
[8] P. Perona and J. Malik, "Scale-Space and Edge Detection Using Anisotropic Diffusion," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 12, no. 7, pp. 629639, July 1990.
[9] A. Witkin, "Scale-Space Filtering," Int'l Joint Conf. Artificial Intelligence, 1983, pp. 1,019-1,021,Karlruhe, West Germany.
[10] G. Barton, Elements of Green's Functions and Propagation,Oxford, England: Clarendon Press, 1989.
[11] P. Perona, T. Shiota, and J. Malik, "Anisotropic Diffusion," Geometry Driven Diffusion in Computer Vision, B.M. Ter Haar Romeny, ed., chapter 3, pp. 73-92. Kluwer, 1994.
[12] L. Alvarez, P.-L. Lions, and J.-M. Morel, "Image Selective Smoothing and Edge Detection by Nonlinear Diffusion, II," SIAM J. Numerical Analysis, vol. 29, no. 3, pp. 845-866, 1992.
[13] L. Alvarez and L. Mazorra, "Signal and Image Restoration Using Shock Filters and Anisotropic Diffusion," SIAM J. Numerical Analysis, vol. 31, no. 2, pp. 590-605, 1994.
[14] F. Catte, P.-L. Lions, J.-M. Morel, and T. Coll, "Image Selective Smoothing and Edge Detection by Nonlinear Diffusion," SIAM J. Numerical Analysis, vol. 29, no. 1, pp. 182-193, 1992.
[15] G.-H. Cottet and L. Germain, "Image Processing Through Reaction Combined with Nonlinear Diffusion," Mathematics of Computation, vol. 61, no. 204, pp. 659-673, 1993.
[16] T. Dang, J. Olivier, and H. Maitre, "An Image Segmentation Technique Based on Edge Preserving Smoothing Filter and Anisotropic Diffusion," Proc. IEEE Southwest Symp. Image Analysis and Interpretation, 1994, pp. 65-69.
[17] A.I. El-Fallah and G.E. Ford, "Nonlinear Adaptive Image Filtering Based on Inhomogenious Diffusion and Differential Geometry," SPIE Image and Video Processing II, vol. 2,182, pp. 49-63, 1994.
[18] B. Engquist, P. Lotstedt, and B. Sjogreen, "Nonlinear Filters for Efficient Shock Computation," Mathematics of Computation, vol. 52, no. 186, pp. 509-537, 1989.
[19] R. Illner and H. Neunzert, "Relative Entropy Maximization and Directed Diffusion Equations," Mathematical Methods in the Applied Sciences, vol. 17, pp. 545-554, 1993.
[20] X. Li and T. Chen, "Nonlinear Diffusion with Multiple Edginess Thresholds," Pattern Recognition, vol. 27, no. 8, pp. 1,029-1,037, 1994.
[21] N.K. Nordstrom, "Biased Anisotropic Diffusion: A Unified Regularization and Diffusion Approach to Edge Detection," Image and Vision Computing, vol. 8, no. 4, pp. 318-327, 1990.
[22] S. Osher and L.I. Rudin, "Feature-Oriented Image Enhancement Using Shock Filters," SIAM J. Numerical Analysis, vol. 27, no. 4, pp. 919-940, 1990.
[23] E.J. Pauwels, M. Proesmans, L.J. Van Gool, T. Moons, and A. Oosterlinck, "Segmentation and Image Enhancement Using Coupled Anisotropic Diffusion," SPIE, vol. 2,094, pp. 836-847, 1993.
[24] S. Fortune,"Polyhedral Modeling with Multiprecision Integer Arithmetic," Proc. Third Symp. on Solid Modeling. ACM Press, 1995.
[25] R.T. Whitaker and S.M. Pizer, "A Multi-Scale Approach to Nonuniform Diffusion," Computer Vision, Graphics and Image Processing, vol. 57, pp. 99-110, 1991.
[26] R.T. Whitaker, "Geometry-Limited Diffusion in the Characterization of Geometric Patches in Images," CVGIP: Image Understanding, vol. 57, no. 1, pp. 111-120, 1993.
[27] J. Kacur and K. Mikula, "Solution of Nonlinear Diffusion Appearing in Image Smoothing and Edge Detection," Applied Numerical Mathematics, vol. 50, pp. 47-59, 1995.
[28] S.J. Thorpe and M. Imbert, "Biological Constraints on Connectionist Models," Connectionism in Perspective, R. Pfeifer, Z. Schreter, and F. Fogelman-Soulie, eds. Amsterdam: Elsevier, 1989, pp. 63-92.
[29] M.W. Oram and D.I. Perrett, "Time Course of Neural Responses Discriminating Different Views of the Face and Head," J. Neurophysiology, vol. 68, no. 1, pp. 70-84, July 1992.
[30] R. Vogels and G.A. Orban, "Quantitative Study of Striate Single Unit Responses in Monkey Performing an Orientation Task," Experimental Brain Research, vol. 84, pp. 1-11, 1991.
[31] R.O. Duda and P.E. Hart, Pattern Classification and Scene Analysis.New York: John Wiley and Sons, 1973.
[32] R. Haberman, Elementary Applied Partial Differential Equations, second edition. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1987.
[33] W. Niessen, B.M. ter Haar Romney, and M. Viergever, "Numerical Analysis of Geometry-Driven Diffusion Equations," Geometry Driven Diffusion in Computer Vision, B.M. Ter Haar Romeny, ed., Computational Imaging and Vision, chapter 15, pp. 393-410. Kluwer, 1994.
[34] R.C. Gonzalez and P. Wintz, Digital Image Processing, second edition. Reading, Mass: Addison-Wesley Publishing Co., 1987.
[35] T.W. Anderson, Introduction to Multivariate Statistics, second edition. New York: John Wiley and Sons, 1971.
[36] P.J. Werbos, "Beyond Regression: New Tools for Prediction and Analysis in the Behavioral Sciences," Ph.D. thesis, Harvard Univ., 1974.
[37] D.B. Parker, "Learning-Logic: Casting the Cortex of the Human Brain in Silicon," Technical Report TR-47, Center for Computational Research in Economics and Management Science, MIT, Cambridge, Mass., 1985.
[38] D.E. Rumelhart, G.E. Hinton, and R.J. Williams, "Learning Representations by Back-Propagating Errors," Nature (London), vol. 323, pp. 533-536, 1986.
[39] Y. LeCun, "Une Procedure D'Apprentissage Pour Reseau A Seuil Assymetrique," Cognitiva, vol. 85, pp. 599-604, 1985.
[40] D.S. Broomhead and D. Lowe, "Multivariable Functional Interpolation and Adaptive Networks," Complex Systems, vol. 2, pp. 321-355, 1988.
[41] J.H. Friedman, "Multivariate Adaptive Regression Splines," The Annals of Statistics, vol. 19, no. 1, pp. 1-50, 1991.
[42] L. Breiman, "TheπMethod for Estimating Multivariate Functions from Noisy Data," Technometrics, vol. 33, no. 2, pp. 125-144, 1991.
[43] C. Blakemore and G.F. Cooper, "Development of the Brain Depends on Visual Environment," Nature (London), vol. 228, pp. 477-478, 1970.
[44] S. Geman and D. Geman, "Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 6, pp. 721-741, 1984.
[45] M.A. Cohen and S. Grossberg, "Some Global Properties of Binocular Resonances: Disparity Matching, Filling-In and Figure-Ground Synthesis," Figural Synthesis, P. Dodwell and T. Caelli, eds. Hillsdale, N.Y.: Erlbaum, 1984.
[46] S. Grossberg and E. Mongilla, "Neural Dynamics of Form Perception: Boundary Completion, Illusory Figures, and Neon Color Spreading," Psychological Rev., vol. 92, pp. 173-211, 1985.
[47] J. Canny, "A Computational Approach to Edge Detection," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 8, pp. 679-698, 1986.

Index Terms:
Anisotropic diffusion, image enhancement, nonlinear filtering, noise reduction.
Bruce Fischl, Eric L. Schwartz, "Learning an Integral Equation Approximation to Nonlinear Anisotropic Diffusion in Image Processing," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 19, no. 4, pp. 342-352, April 1997, doi:10.1109/34.588012
Usage of this product signifies your acceptance of the Terms of Use.