Robust Hypersurface Fitting Based on Random Sampling Approximations

This paper considers N−1-dimensional hypersurface fitting based on L2 distance in N-dimensional input space. The problem is usually reduced to hyperplane fitting in higher dimension. However, because feature mapping is generally a nonlinear mapping, it does not preserve the order of lengthes, and this derives an unacceptable fitting result. To avoid it, JNLPCA is introduced. JNLPCA defines the L2 distance in the feature space as a weighted L2 distance to reflect the metric in the input space. In the fitting, random sampling approximation of least k-th power deviation, and least α-percentile of squares are introduced to make estimation robust. The proposed hypersurface fitting method is evaluated by quadratic curve fitting and quadratic curve segments extraction from artificial data and a real image.

[1]  Robert C. Bolles,et al.  Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography , 1981, CACM.

[2]  Peter J. Rousseeuw,et al.  Robust regression and outlier detection , 1987 .

[3]  Jun Fujiki,et al.  Robust Hyperplane Fitting Based on k-th Power Deviation and α-Quantile , 2011, CAIP.

[4]  Jun Fujiki,et al.  Hypersurface Fitting via Jacobian Nonlinear PCA on Riemannian Space , 2011, CAIP.

[5]  Gabriel Taubin,et al.  Estimation of Planar Curves, Surfaces, and Nonplanar Space Curves Defined by Implicit Equations with Applications to Edge and Range Image Segmentation , 1991, IEEE Trans. Pattern Anal. Mach. Intell..

[6]  Wojciech Chojnacki,et al.  On the Fitting of Surfaces to Data with Covariances , 2000, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  Kenichi Kanatani,et al.  Unified Computation of Strict Maximum Likelihood for Geometric Fitting , 2010, Journal of Mathematical Imaging and Vision.

[8]  Richard O. Duda,et al.  Use of the Hough transformation to detect lines and curves in pictures , 1972, CACM.

[9]  Shotaro Akaho,et al.  Curve fitting that minimizes the mean square of perpendicular distances from sample points , 1993, Other Conferences.

[10]  PAUL D. SAMPSON,et al.  Fitting conic sections to "very scattered" data: An iterative refinement of the bookstein algorithm , 1982, Comput. Graph. Image Process..

[11]  Peter J. Rousseeuw,et al.  Robust Regression and Outlier Detection , 2005, Wiley Series in Probability and Statistics.

[12]  Erkki Oja,et al.  A new curve detection method: Randomized Hough transform (RHT) , 1990, Pattern Recognit. Lett..