Abstract
The focus of this paper is on the mathematical construction of a transform that is invariant to a finite rotation group and is stable to small perturbations. A key step in our theory lies in the construction of directionally sensitive functions that are partially generated by rotations. We call such a family a rotational uniform covering frame and by studying rotations of the frame, we derive the desired operator, which we call the rotational Fourier scattering transform. We prove that the transformation is rotationally invariant to a finite rotation group, is bounded above and below, is non-expansive, and contracts small translations and additive diffeomorphisms. To address the numerical aspects of this theory, we also construct digital versions of the frame and show how to faithfully truncate the transform. We also discuss connections between this new family of directional representations with previously constructed ones.
Similar content being viewed by others
References
Andén, J., Mallat, S.: Deep scattering spectrum. IEEE Trans. Signal Process. 62(16), 4114–4128 (2014)
Arivazhagan, S., Ganesan, L., Padam Priyal, S.: Texture classification using Gabor wavelets based rotation invariant features. Pattern Recognit. Lett. 27(16), 1976–1982 (2006)
Bammer, R., Dörfler, M. : Invariance and stability of Gabor scattering for music signals. In: 2017 International Conference on Sampling Theory and Applications (SampTA), pp. 299–302. IEEE (2017)
Bosch, E.H., Czaja, W., Murphy, J.M., Weinberg, D.: Anisotropic representations for superresolution of hyperspectral data. In: Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XXI, vol. 9472, pp. 947213. International Society for Optics and Photonics (2015)
Bruna, J., Mallat, S.: Invariant scattering convolution networks. IEEE Trans. Pattern Anal. Mach. Intell. 35(8), 1872–1886 (2013)
Candès, E.J.: Harmonic analysis of neural networks. Appl. Comput. Harmon. Anal. 6(2), 197–218 (1999)
Candès, E.J., Donoho, D.L.: New tight frames of curvelets and optimal representations of objects with piecewise \(c^2\) singularities. Commun. Pure Appl. Math. 57(2), 219–266 (2004)
Cheng, X., Chen, X., Mallat, S.: Deep Haar scattering networks. Inf. Inference 5(2), 105–133 (2016)
Czaja, W., Kavalerov, I., Li, W.: Scattering transforms and classification of hyperspectral images. In Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XXIV, vol. 10644, p. 106440H. International Society for Optics and Photonics (2018)
Czaja, W., King, E.J.: Isotropic shearlet analogs for \({L}^2(\mathbb{R}^k)\) and localization operators. Numer. Funct. Anal. Optim. 33(7–9), 872–905 (2012)
Czaja, W., King, E.J.: Anisotropic shearlet transforms for \({L}^2(\mathbb{R}^k)\). Math. Nachr. 287(8–9), 903–916 (2014)
Czaja, W., Li, W.: Analysis of time–frequency scattering transforms. Appl. Comput. Harmon. Anal. 47(1), 149–171 (2019)
Czaja, W., Manning, B., Murphy, J.M., Stubbs, K.: Discrete directional Gabor frames. Appl. Comput. Harmon. Anal. 45(1), 1–21 (2018)
Grafakos, L., Sansing, C.: Gabor frames and directional time-frequency analysis. Appl. Comput. Harmon. Anal. 25(1), 47–67 (2008)
Guo, K., Labate, D.: Optimally sparse multidimensional representation using shearlets. SIAM J. Math. Anal. 39(1), 298–318 (2007)
Hamamoto, Y., Uchimura, S., Watanabe, M., Yasuda, T., Mitani, Y., Tomita, S.: A Gabor filter-based method for recognizing handwritten numerals. Pattern Recognit. 31(4), 395–400 (1998)
Hernández, Eugenio, Weiss, Guido: A First Course on Wavelets. CRC Press, Cambridge (1996)
Hirn, M., Mallat, S., Poilvert, N.: Wavelet scattering regression of quantum chemical energies. Multiscale Model. Simul. 15(2), 827–863 (2017)
Kavalerov, I., Li, W., Czaja, W., Chellappa, R.: Three-dimensional Fourier scattering transform and classification of hyperspectral images. arXiv preprint arXiv:1906.06804 (2019)
Kong, W.K., Zhang, D., Li, W.: Palmprint feature extraction using 2-D Gabor filters. Pattern Recognit. 36(10), 2339–2347 (2003)
Labate, D., Lim, W.-Q., Kutyniok, G., Weiss, G. : Sparse multidimensional representation using shearlets. In: Wavelets XI, vol. 5914, p. 59140U. International Society for Optics and Photonics (2005)
Li, W.: Topics in Harmonic Analysis, Sparse Representations, and Data Analysis. PhD thesis, University of Maryland, College Park (2018)
Mallat, S.: Group invariant scattering. Commun. Pure Appl. Math. 65(10), 1331–1398 (2012)
Mallat, S.: Understanding deep convolutional networks. Philos. Trans. R. Soc. A 374(2065), 20150203 (2016)
Ojala, T., Pietikäinen, M., Mäenpää, T.: Multiresolution gray-scale and rotation invariant texture classification with local binary patterns. IEEE Trans. Pattern Anal. Mach. Intell. 24(7), 971–987 (2002)
Rowley, H.A., Baluja, S., Kanade, T: Rotation invariant neural network-based face detection. In: Proceedings of the IEEE computer society conference on computer vision and pattern recognition, p. 38. IEEE Computer Society (1998)
Sifre, L., Mallat, S.: Rotation, scaling and deformation invariant scattering for texture discrimination. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 1233–1240 (2013)
Stein, E.M.: Harmonic Analysis: Real-Variable Methods, Orthogonality, and Oscillatory Integrals, vol. 3. Princeton University Press, Princeton (1993)
Waldspurger, I.: Exponential decay of scattering coefficients. In: 2017 international conference on sampling theory and applications (SampTA), pp. 143–146. IEEE (2017)
Wiatowski, T., Bölcskei, H.: A mathematical theory of deep convolutional neural networks for feature extraction. IEEE Trans. Inf. Theory 64(3), 1845–1866 (2018)
Zou, D., Lerman, G.: Graph convolutional neural networks via scattering. Appl. Comput. Harmon. Anal. https://doi.org/10.1016/j.acha.2019.06.003 (2019)
Acknowledgements
This work was supported in part by the LTS through Maryland Procurement Office, NSF DMS grant 1738003, Defense Threat Reduction Agency grant HDTRA 1-13-1-0015, and by the Army Research Office Grants W911 NF-15-1-0112 and W911 NF-16-1-0008. Weilin Li was supported by the National Science Foundation Grant DMS-1440140 while in residence at the Mathematical Sciences Research Institute in Berkeley, California, during the Spring 2017 semester.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Thomas Strohmer.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Czaja, W., Li, W. Rotationally Invariant Time–Frequency Scattering Transforms. J Fourier Anal Appl 26, 4 (2020). https://doi.org/10.1007/s00041-019-09705-w
Received:
Revised:
Published:
DOI: https://doi.org/10.1007/s00041-019-09705-w
Keywords
- Scattering transform
- Uniform covering frames
- Time–frequency
- Directional representations
- Neural networks
- Feature extraction