Abstract
In this and the following chapters, we demonstrate why multiple subspaces can be a very useful class of models for image processing and how the subspace clustering techniques may facilitate many important image processing tasks, such as image representation, compression, image segmentation, and video segmentation.
Everything should be made as simple as possible, but not simpler.
—Albert Einstein
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Which involves further quantization and entropy coding of the so-obtained sparse signals α.
- 2.
A topic we will study in more detail from the compression perspective in the next section.
- 3.
Here, “optimality” means that the transformation achieves the optimal asymptotic for approximating the class of functions considered (DeVore 1998).
- 4.
Here, in contrast to the case of pre-fixed transformations, “optimality” means the representation obtained is the optimal one within the class of models considered, in the sense that it minimizes certain discrepancies between the model and the data.
- 5.
In fact, in the VQ model, the coefficients are assumed to be binary.
- 6.
Be aware that compared to methods in the first category, representations in this category typically need additional memory to store the information about the resulting model itself, e.g., the basis of the subspace in PCA, the cluster means in VQ.
- 7.
Therefore, b needs to be a common divisor of W and H.
- 8.
Here by default, the peak value of the imagery data is normalized to 1.
- 9.
Notice that to represent a d-dimensional subspace in a D-dimensional space, we need only specify a basis of d linearly independent vectors for the subspace. We may stack these vectors as rows of a \(d \times D\) matrix. Any nonsingular linear transformation of these vectors span the same subspace. Thus, without loss of generality, we may assume that the matrix is of the normal form \([I_{d\times d},G]\), where G is a \(d \times (D - d)\) matrix consisting of the so-called Grassmannian coordinates.
- 10.
Notice that if one uses a preselected basis, such as discrete Fourier transform, discrete cosine transform (JPEG), or wavelets (JPEG-2000), there is no such overhead.
- 11.
We also need a very small number of binary bits to store the membership of the vectors. But those extra bits are insignificant compared to \(\Omega \) and often can be ignored.
- 12.
In fact, the minimal \(\Omega \) can also be associated with the Kolmogorov entropy or with the minimum description length (MDL) of the imagery data.
- 13.
For instance, if we take all the \(b \times b\) blocks and scramble them arbitrarily, the scrambled image would be fit equally well by the same hybrid linear model for the original image.
- 14.
This is not to be confused with the subscript i used to indicate different segments \(\mathcal{I}_{i}\) of an image.
References
Bruckstein, A., Donoho, D., & Elad, M. (2009). From sparse solutions of systems of equations to sparse modeling of signals and images. SIAM Review, 51(1), 34–81.
Burt, P. J., & Adelson, E. H. (1983). The Laplacian pyramid as a compact image code. IEEE Transactions on Communications, 31(4), 532–540.
Candès, E. (2006). Compressive sampling. In Proceedings of the International Congress of Mathematics.
Candès, E., & Donoho, D. (2002). New tight frames of curvelets and optimal representations of objects with smooth singularities. Technical Report. Stanford University.
Candès, E., & Wakin, M. (2008). An introduction to compressive sampling. IEEE Signal Processing Magazine, 25(2), 21–30.
Chen, J.-Q., Pappas, T. N., Mojsilovic, A., & Rogowitz, B. E. (2003). Image segmentation by spatially adaptive color and texture features. In IEEE International Conference on Image Processing.
Chen, S., Donoho, D., & Saunders, M. (1998). Atomic decomposition by basis pursuit. SIAM Journal of Scientific Computing, 20(1), 33–61.
Coifman, R., & Wickerhauser, M. (1992). Entropy-based algorithms for best bases selection. IEEE Transactions on Information Theory, 38(2), 713–718.
Delsarte, P., Macq, B., & Slock, D. (1992). Signal-adapted multiresolution transform for image coding. IEEE Transactions on Information Theory, 38, 897–903.
DeVore, R. (1998). Nonlinear approximation. Acta Numerica, 7, 51–150.
DeVore, R., Jawerth, B., & Lucier, B. (1992). Image compression through wavelet transform coding. IEEE Transactions on Information Theory, 38(2), 719–746.
Do, M. N., & Vetterli, M. (2002). Contourlets: A directional multiresolution image representation. In IEEE International Conference on Image Processing.
Donoho, D. (1995). Cart and best-ortho-basis: A connection. Manuscript.
Donoho, D. (1998). Sparse components analysis and optimal atomic decomposition. Technical Report, Department of Statistics, Stanford University.
Donoho, D. L. (1999). Wedgelets: Nearly-minimax estimation of edges. Annals of Statistics, 27, 859–897.
Donoho, D. L., & Elad, M. (2003). Optimally sparse representation in general (nonorthogonal) dictionaries via ℓ 1 minimization. Proceedings of National Academy of Sciences, 100(5), 2197–2202.
Donoho, D. L., Vetterli, M., DeVore, R., & Daubechies, I. (1998). Data compression and harmonic analysis. IEEE Transactions on Information Theory, 44(6), 2435–2476.
Effros, M., & Chou, P. (1995). Weighted universal transform coding: Universal image compression with the Karhunen-Loéve transform. In IEEE International Conference on Image Processing (Vol. 2, pp. 61–64).
Elad, M., & Bruckstein, A. (2001). On sparse signal representations. In IEEE International Conference on Image Processing.
Elad, M., & Bruckstein, A. (2002). A generalized uncertainty principle and sparse representation in pairs of bases. IEEE Transactions on Information Theory, 48(9), 2558–2567.
Elad, M., Figueiredo, M. A. T., & Ma, Y. (2010). On the role of sparse and redundant representations in image processing. Proceedings of the IEEE, 98(6), 972–982.
Feuer, A., Nemirovski, A. (2003). On sparse representation in pairs of bases. IEEE Transactions on Information Theory, 49(6), 1579–1581.
Fisher, Y. (1995). Fractal Image Compression: Theory and Application. Springer-Verlag Telos.
Gersho, A., & Gray, R. M. (1992). Vector Quantization and Signal Compression. Boston: Kluwer Academic.
Hong, W., Wright, J., Huang, K., & Ma, Y. (2006). Multi-scale hybrid linear models for lossy image representation. IEEE Transactions on Image Processing, 15(12), 3655–3671.
Hotelling, H. (1933). Analysis of a complex of statistical variables into principal components. Journal of Educational Psychology, 24, 417–441.
Jolliffe, I. (2002). Principal Component Analysis (2nd ed.). New York: Springer.
LePennec, E., & Mallat, S. (2005). Sparse geometric image representation with bandelets. IEEE Transactions on Image Processing, 14(4), 423–438.
Mallat, S. (1999). A Wavelet Tour of Signal Processing (2nd ed.). London: Academic.
Meyer, F. (2000). Fast adaptive wavelet packet image compression. IEEE Transactions on Image Processing, 9(5), 792–800.
Meyer, F. (2002). Image compression with adaptive local cosines. IEEE Transactions on Image Processing, 11(6), 616–629.
Muresan, D., & Parks, T. (2003). Adaptive principal components and image denoising. In IEEE International Conference on Image Processing.
Olshausen, B., & D.J.Field (1996). Emergence of simple-cell receptive field properties by learning a sparse code for natural images. Nature, 381(6583), 607–609.
Pavlovic, V., Moulin, P., & Ramchandran, K. (1998). An integrated framework for adaptive subband image coding. IEEE Transactions on Signal Processing, 47(4), 1024–1038.
Pearson, K. (1901). On lines and planes of closest fit to systems of points in space. The London, Edinburgh and Dublin Philosphical Magazine and Journal of Science, 2, 559–572.
Rabiee, H., Kashyap, R., & Safavian, S. (1996). Adaptive multiresolution image coding with matching and basis pursuits. In IEEE International Conference on Image Processing.
Ramchandran, K., & Vetterli, M. (1993). Best wavelet packets bases in a rate-distortion sense. IEEE Transactions on Image Processing, 2, 160–175.
Ramchandran, K., Vetterli, M., & Herley, C. (1996). Wavelets, subband coding, and best basis. Proceedings of the IEEE, 84(4), 541–560.
Shapiro, J. M. (1993). Embedded image coding using zerotrees of wavelet coefficients. IEEE Transactions on Signal Processing, 41(12), 3445–3463.
Sikora, T., & Makai, B. (1995). Shape-adaptive DCT for generic coding of video. IEEE Transactions on Circuits and Systems For Video Technology, 5, 59–62.
Spielman, D., Wang, H., & Wright, J. (2012). Exact recovery of sparsity-used dictionaries. Conference on Learning Theory (COLT).
Starck, J.-L., Elad, M., & Donoho, D. (2003). Image decomposition: Separation of texture from piecewise smooth content. In Proceedings of the SPIE (Vol. 5207, pp. 571–582).
Sun, J., Qu, Q., & Wright, J. (2015). Complete dictionary recovery over the sphere. Preprint. http://arxiv.org/abs/1504.06785
Vetterli, M., & Kovacevic, J. (1995). Wavelets and subband coding. Upper Saddle River: Prentice-Hall.
Wallace, G. K. (1991). The JPEG still picture compression standard. Communications of the ACM. Special issue on digital multimedia systems, 34(4), 30–44.
Yang, J., Wright, J., Huang, T., & Ma, Y. (2010). Image super-resolution via sparse representation. IEEE Transactions on Image Processing, 19(11), 2861–2873.
Yu, G., Sapiro, G., & Mallat, S. (2010). Image modeling and enhancement via structured sparse model selection. In International Conference on Image Processing.
Yu, G., Sapiro, G., & Mallat, S. (2012). Solving inverse problems with piecewise linear estimators: From gaussian mixture models to structured sparsity. IEEE Transactions on Image Processing, 21(5), 2481–2499.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2016 Springer-Verlag New York
About this chapter
Cite this chapter
Vidal, R., Ma, Y., Sastry, S.S. (2016). Image Representation. In: Generalized Principal Component Analysis. Interdisciplinary Applied Mathematics, vol 40. Springer, New York, NY. https://doi.org/10.1007/978-0-387-87811-9_9
Download citation
DOI: https://doi.org/10.1007/978-0-387-87811-9_9
Published:
Publisher Name: Springer, New York, NY
Print ISBN: 978-0-387-87810-2
Online ISBN: 978-0-387-87811-9
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)