On the choice of consistent canonical form during moment normalization

The key issue in normalization with respect to geometric transformation is how to obtain a consistent canonical form which remains unchanged for different geometric transformation related images. Due to ambiguities inherent in the specified normalization methods, more than one canonical form may occur during the normalization procedure. This causes difficulties in obtaining the expected invariant features or the transformation parameters through normalization. This paper aims to provide general schemes to analyze the ambiguity characteristics and to derive the ambiguity matrices through which ambiguities can be eliminated. Three kinds of ambiguities caused by the multi-roots of high-order polynomials, the symmetrical normalization constraints, and the reflection are addressed and solutions are provided to obtain a consistent canonical form for each kind of ambiguities.

[1]  Herbert Süße,et al.  The Method of Normalization to Determine Invariants , 1996, IEEE Trans. Pattern Anal. Mach. Intell..

[2]  Soo-Chang Pei,et al.  Image normalization for pattern recognition , 1995, Image Vis. Comput..

[3]  Anthony P. Reeves,et al.  Three-Dimensional Shape Analysis Using Moments and Fourier Descriptors , 1988, IEEE Trans. Pattern Anal. Mach. Intell..

[4]  Herbert Süße,et al.  Invariant Fitting of Planar Objects by Primitives , 1997, IEEE Trans. Pattern Anal. Mach. Intell..

[5]  Jian Wu,et al.  Estimiting the finite displacement using moments , 1990, Pattern Recognit. Lett..

[6]  Dinggang Shen,et al.  Generalized Affine Invariant Image Normalization , 1997, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  Ying Zhang,et al.  Determination of blur and affine combined invariants by normalization , 2002, Pattern Recognit..

[8]  Andrew Blake,et al.  Isoperimetric Normalization of Planar Curves , 1994, IEEE Trans. Pattern Anal. Mach. Intell..

[9]  Georgios Tziritas,et al.  Motion estimation based on affine moment invariants , 1998, 9th European Signal Processing Conference (EUSIPCO 1998).

[10]  Jia-Guu Leu,et al.  Shape normalization through compacting , 1989, Pattern Recognit. Lett..

[11]  Ying Zhang,et al.  Estimation of motion parameters from blurred images , 2000, Pattern Recognit. Lett..

[12]  Yannis Avrithis,et al.  Affine-invariant curve normalization for shape-based retrieval , 2000, Proceedings 15th International Conference on Pattern Recognition. ICPR-2000.

[13]  Stephen L. Adler,et al.  Similarity And Affine Normalization Of Partially Occluded Planar Curves Using First And Second Derivatives , 1998, Pattern Recognit..

[14]  Thomas H. Reiss,et al.  Recognizing Planar Objects Using Invariant Image Features , 1993, Lecture Notes in Computer Science.