Theory of Probability and Mathematical Statistics
Full inference for the anisotropic fractional Brownian field
Paul Escande and Frédéric J. P. Richard
Link
Abstract: The anisotropic fractional Brownian field (AFBF) is a non-stationary Gaussian random field which has been used for the modeling of textured images. In this paper, we address the open issue of estimating the functional parameters of this field, namely the topothesy and Hurst functions. We propose an original method which fits the empirical semi-variogram of an image to the semi-variogram of a turning-band field that approximates the AFBF. Expressing the fitting criterion in terms of a separable non-linear least square criterion, we design a minimization algorithm inspired by the variable projection approach. This algorithm also includes a coarse-to-fine multigrid strategy based on approximations of functional parameters. Compared to existing methods, the new method enables to estimate both functional parameters on their whole definition domain. On simulated textures, we show that it has a low estimation error, even when the parameters are approximated with a high precision. We also apply the method to characterize mammograms and sample images with synthetic parenchymal patterns.
Keywords: Inference from random field, anisotropic fractional Brownian fields, image texture analysis, image texture synthesis
Bibliography: A. Arneodo, N. Decoster, P. Kestener, and S. G. Roux, A wavelet-based multifractal image analysis: From theoretical concepts to experimental applications, Adv. Imaging Electron Phys. 126 (2003), 1–92.
P. Bakic, M. Albert, D. Brzakovic, and A. Maidment, Mammogram synthesis using a 3D simulation. I. Breast tissue model and image acquisition simulation, Med. Phys. 29 (2002), 2131–2139.
—, Mammogram synthesis using a 3D simulation. II. Evaluation of synthetic mammogram texture, Med. Phys. 29 (2002), no. 9, 2140–2151.
—, Mammogram synthesis using a 3D simulation. III. Modeling and evaluation of the breast ductal network, Med. Phys. 30 (2003), no. 7, 1914–1925.
Albert Benassi, Stéphane Jaffard, and Daniel Roux, Elliptic Gaussian random processes, Rev. Mat. Iberoamericana 13 (1997), no. 1, 19–90 (English, with English and French summaries). MR 1462329, DOI 10.4171/RMI/217
Hermine Biermé, Lionel Moisan, and Frédéric Richard, A turning-band method for the simulation of anisotropic fractional Brownian fields, J. Comput. Graph. Statist. 24 (2015), no. 3, 885–904. MR 3397238, DOI 10.1080/10618600.2014.946603
Hermine Biermé and Frédéric Richard, Estimation of anisotropic Gaussian fields through Radon transform, ESAIM Probab. Stat. 12 (2008), 30–50. MR 2367992, DOI 10.1051/ps:2007031
Hermine Biermé and Frédéric J. P. Richard, Analysis of texture anisotropy based on some Gaussian fields with spectral density, Mathematical image processing, Springer Proc. Math., vol. 5, Springer, Heidelberg, 2011, pp. 59–73. MR 2867520, DOI 10.1007/978-3-642-19604-1_{3}
Hermine Biermé, Frédéric Richard, Mouna Rachidi, and Claude-Laurent Benhamou, Anisotropic texture modeling and applications to medical image analysis, Mathematical methods for imaging and inverse problems, ESAIM Proc., vol. 26, EDP Sci., Les Ulis, 2009, pp. 100–122 (English, with English and French summaries). MR 2498142, DOI 10.1051/proc/2009008
Aline Bonami and Anne Estrade, Anisotropic analysis of some Gaussian models, J. Fourier Anal. Appl. 9 (2003), no. 3, 215–236. MR 1988750, DOI 10.1007/s00041-003-0012-2
Mary Ann Branch, Thomas F. Coleman, and Yuying Li, A subspace, interior, and conjugate gradient method for large-scale bound-constrained minimization problems, SIAM J. Sci. Comput. 21 (1999), no. 1, 1–23. MR 1722103, DOI 10.1137/S1064827595289108
A. Burgess, F. Jacobson, and P. Judy, Human observer detection experiments with mammograms and power-law noise, Med. Phys. 28 (2001), no. 4, 419–437.
A.-K. Carton, P. Bakic, C. Ullberg, H. Derand, and A. Maidment, Development of a physical 3D anthropomorphic breast phantom, Med. Phys. 38 (2011), no. 2, 891–896.
Bruno Galerne, Yann Gousseau, and Jean-Michel Morel, Random phase textures: theory and synthesis, IEEE Trans. Image Process. 20 (2011), no. 1, 257–267. MR 2789729, DOI 10.1109/TIP.2010.2052822
B. Galerne, A. Leclaire, and L. Moisan, A texton for fast and flexible Gaussian texture synthesis, 2014 22nd European Signal Processing Conference (EUSIPCO), IEEE, 2014, 1686–1690.
G. H. Golub and V. Pereyra, The differentiation of pseudo-inverses and nonlinear least squares problems whose variables separate, SIAM J. Numer. Anal. 10 (1973), 413–432. MR 336980, DOI 10.1137/0710036
—, Separable nonlinear least squares: The variable projection method and its applications, Inverse Probl. 19 (2003), no. 2, R1.
J. Heine, S. Deine, R. Velthuizen, and L. P. Clarke, On the statistical nature of mammograms, Med. Phys. 26 (1999), no. 11, 2254–2269.
J. Heine and R. Velthuizen, Spectral analysis of full field digital mammography data, Med. Phys. 29 (2002), no. 5, 647–661.
Linda Kaufman, A variable projection method for solving separable nonlinear least squares problems, Nordisk Tidskr. Informationsbehandling (BIT) 15 (1975), no. 1, 49–57. MR 501738, DOI 10.1007/bf01932995
P. Kestener, J.-M. Lina, P. Saint-Jean, and A. Arneodo, Wavelet-based multifractal formalism to assist in diagnosis in digitized mammograms, Image Anal. Stereol. 20 (2001), 169–174.
Kenneth Levenberg, A method for the solution of certain non-linear problems in least squares, Quart. Appl. Math. 2 (1944), 164–168. MR 10666, DOI 10.1090/S0033-569X-1944-10666-0
Donald W. Marquardt, An algorithm for least-squares estimation of nonlinear parameters, J. Soc. Indust. Appl. Math. 11 (1963), 431–441. MR 153071, DOI 10.1137/0111030
Dianne P. O’Leary and Bert W. Rust, Variable projection for nonlinear least squares problems, Comput. Optim. Appl. 54 (2013), no. 3, 579–593. MR 3029309, DOI 10.1007/s10589-012-9492-9
Barbara Pascal, Nelly Pustelnik, and Patrice Abry, Strongly convex optimization for joint fractal feature estimation and texture segmentation, Appl. Comput. Harmon. Anal. 54 (2021), 303–322. MR 4241986, DOI 10.1016/j.acha.2021.03.009
Barbara Pascal, Samuel Vaiter, Nelly Pustelnik, and Patrice Abry, Automated data-driven selection of the hyperparameters for total-variation-based texture segmentation, J. Math. Imaging Vision 63 (2021), no. 7, 923–952. MR 4293954, DOI 10.1007/s10851-021-01035-1
K. Polisano, M. Clausel, V. Perrier, and L. Condat, Texture modeling by Gaussian fields with prescribed local orientation, Int Conf on Image Processing (ICIP), 2014 IEEE, 2014, 6091–6095.
Frédéric J. P. Richard, Anisotropy of Hölder Gaussian random fields: characterization, estimation, and application to image textures, Stat. Comput. 28 (2018), no. 6, 1155–1168. MR 3850388, DOI 10.1007/s11222-017-9785-z
—, Analysis of anisotropic Brownian textures and application to lesion detection in mammograms, Procedia Environ. Sci. 27 (2015), 16–20.
Frédéric J. P. Richard, Some anisotropy indices for the characterization of Brownian textures and their application to breast images. part A, Spat. Stat. 18 (2016), no. part A, 147–162. MR 3573274, DOI 10.1016/j.spasta.2016.02.001
Frédéric J. P. Richard, Tests of isotropy for rough textures of trended images, Statist. Sinica 26 (2016), no. 3, 1279–1304. MR 3559953
—, PyAFBF: A Python library for sampling image textures from the anisotropic fractional Brownian field., J. Open Source Softw. 7 (2022), no. 75, 3821.
F.J.P. Richard and H. Biermé, Statistical tests of anisotropy for fractional Brownian textures. Application to full-field digital mammography, J. Math. Imaging Vis. 36 (2010), no. 3, 227–240.
Axel Ruhe and Per Ȧke Wedin, Algorithms for separable nonlinear least squares problems, SIAM Rev. 22 (1980), no. 3, 318–337. MR 584380, DOI 10.1137/1022057
C. Shorten and T. Khoshgoftaar, A survey on image data augmentation for deep learning, J. Big Data 6 (2019), no. 1, 1–48.
Tristan van Leeuwen and Aleksandr Y. Aravkin, Variable projection for nonsmooth problems, SIAM J. Sci. Comput. 43 (2021), no. 5, S249–S268. MR 4331946, DOI 10.1137/20M1348650
P. Virtanen, R. Gommers, T. E. Oliphant, et al., SciPy 1.0: Fundamental algorithms for scientific computing in Python, Nat. Methods 17 (2020), 261–272.
Huong T. L. Vu and Frédéric J. P. Richard, Statistical tests of heterogeneity for anisotropic multifractional Brownian fields, Stochastic Process. Appl. 130 (2020), no. 8, 4667–4692. MR 4108467, DOI 10.1016/j.spa.2020.01.012
Yu Wang, Wotao Yin, and Jinshan Zeng, Global convergence of ADMM in nonconvex nonsmooth optimization, J. Sci. Comput. 78 (2019), no. 1, 29–63. MR 3902876, DOI 10.1007/s10915-018-0757-z
