Optimal template matching by nonorthogonal image expansion using restoration

In this paper we present a novel approach for template matching. The basic principle is expansion matching and it entails signal expansion into a set of nonorthogonal templatesimilar basis functions. The coefficients of this expansion signify the presence of the template in the corresponding locations in the image. We demonstrate that this matching technique is robust in conditions of noise, superposition, and severe occlusion. A new and more practical discriminative signal-to-noise ratio (DSNR) for matching is proposed that considers even the filter's off-center response to the template as “noise”. We show that expansion yields the optimal linear operator that maximizes the DSNR and results in a sharp response to the matched template. Theoretical and experimental comparisons of expansion matching and the widely used correlation matching demonstrate the superiority of our approach. Correlation matching (also known as matched filtering) yields broad peaks and spurious responses, both of which hamper good detection. We also show that the special case of expansion with a dense set of self-similar basis functions is equivalent to signal restoration. Expansion matching can be implemented by restoration techniques and also by our recently developed lattice architecture.

[1]  J. L. Harris.  Resolving Power and Decision Theory , 1964 .

[2]  Alex Pentland,et al.  Face recognition using eigenfaces , 1991, Proceedings. 1991 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[3]  K.R. Rao,et al.  Signal representation by generalized nonorthogonal Gaussian wavelet groups using lattice networks , 1991, [Proceedings] 1991 IEEE International Joint Conference on Neural Networks.

[4]  Werner Frei,et al.  Fast Boundary Detection: A Generalization and a New Algorithm , 1977, IEEE Transactions on Computers.

[5]  Jezekiel Ben-Arie,et al.  On the Use of Non-Orthogonal Signal Expansions for Recognitlon , 1992, 1992 American Control Conference.

[6]  Touradj Ebrahimi,et al.  Image compression by Gabor expansion , 1991 .

[7]  T. M. Cannon,et al.  Blind deconvolution through digital signal processing , 1975, Proceedings of the IEEE.

[8]  B. V. Vijaya Kumar,et al.  Minimum-variance synthetic discriminant functions , 1986 .

[9]  P. W. Jones,et al.  Digital Image Compression Techniques , 1991 .

[10]  Jezekiel Ben-Arie,et al.  Generic face recognition, feature extraction and edge detection using optimal DSNR expansion matching , 1993, 1993 IEEE International Symposium on Circuits and Systems.

[11]  C. Helstrom.  Image Restoration by the Method of Least Squares , 1967 .

[12]  John Daugman,et al.  Neural networks for image transformation, analysis, and compression , 1988, Neural Networks.

[13]  Manfred H. Hueckel.  An Operator Which Locates Edges in Digitized Pictures , 1971, J. ACM.

[14]  K.R. Rao,et al.  Parallel Generation Of Fourier And Gabor Transforms And Shape Descriptors By Gaussian Wavelet Groups Using A Set Of Multidimensional Lattices , 1991, Proceedings of the Seventh Workshop on Multidimensional Signal Processing.

[15]  Jezekiel Ben-Arie.  Multi-Dimensional Linear Lattice for Fourier and Gabor Transforms, Multiple-Scale Gaussian Filtering, and Edge Detection , 1992 .

[16]  K.R. Rao,et al.  Lattice architectures for signal expansion by Gaussian set wavelets with applications to recognition , 1992, [Proceedings] 1992 IEEE International Symposium on Circuits and Systems.

[17]  G.L. Turin,et al.  An introduction to digitial matched filters , 1976, Proceedings of the IEEE.

[18]  Robert A. Hummel,et al.  Feature detection using basis functions , 1979 .

[19]  Athanasios Papoulis,et al.  Probability, Random Variables and Stochastic Processes , 1965 .

[20]  Jezekiel Ben-Arie,et al.  Linear lattice architectures that utilize the central limit for image analysis, Gaussian operators, sine, cosine, Fourier, and Gabor transforms , 1991, Other Conferences.

[21]  George L. Turin,et al.  An introduction to digital matched filters , 1976 .

[22]  D Casasent,et al.  Unified synthetic discriminant function computational formulation. , 1984, Applied optics.

[23]  K.R. Rao,et al.  A lattice network for signal representation using Gaussian basis functions and max-energy paradigm , 1991, [1991] Proceedings of the 34th Midwest Symposium on Circuits and Systems.

[24]  J.F. Andrus,et al.  Digital Image Registration Method Using Boundary Maps , 1975, IEEE Transactions on Computers.

[25]  Robert J. Schalkoff,et al.  Digital Image Processing and Computer Vision , 1989 .

[26]  B. Roy Frieden,et al.  Band-Unlimited Reconstruction of Optical Objects and Spectra* , 1967 .

[27]  B. Frieden.  Restoring with maximum likelihood and maximum entropy. , 1972, Journal of the Optical Society of America.

[28]  M. Horowitz.  Efficient Use of a Picture Correlator , 1957 .

[29]  J. R. Higgins.  Completeness and basis properties of sets of special functions: Bibliography , 1977 .

[30]  D. Casasent,et al.  Minimum average correlation energy filters. , 1987, Applied optics.

[31]  K. Raghunath Rao,et al.  A novel approach for template matching by nonorthogonal image expansion , 1993, IEEE Trans. Circuits Syst. Video Technol..

[32]  H. Saunders,et al.  Probability, Random Variables and Stochastic Processes (2nd Edition) , 1989 .

[33]  Jezekiel Ben-Arie,et al.  Optimal operators for template and shape recognition by nonorthogonal image expansion , 1992, Other Conferences.

[34]  Martin D. Levine,et al.  Vision in Man and Machine , 1985 .

[35]  Haddan Mostafavier,et al.  Image Correlation with Geometric Distortion Part 1: Acquisition Performance , 1978, IEEE Transactions on Aerospace and Electronic Systems.

[36]  K. Raghunath Rao,et al.  Image expansion by non-orthogonal basis functions extended for optimal multiple template matching , 1993, 1993 IEEE International Conference on Acoustics, Speech, and Signal Processing.