Abstract
In its most widespread imaging and vision applications, Ambrosio and Tortorelli (AT) phase field is a technical device for applying gradient descent to Mumford and Shah simultaneous segmentation and restoration functional or its extensions. As such, it forms a diffuse alternative to sharp interfaces or level sets and parametric techniques. The functionality of the AT field, however, is not limited to segmentation and restoration applications. We demonstrate the possibility of coding parts—features that are higher level than edges and boundaries—after incorporating higher level influences via distances and averages. The iteratively extracted parts using the level curves with double point singularities are organized as a proper binary tree. Inconsistencies due to non-generic configurations for level curves as well as due to visual changes such as occlusion are successfully handled once the tree is endowed with a probabilistic structure. As a proof of concept, we present (1) the most probable configurations from our randomized trees; and (2) correspondence matching results between illustrative shape pairs.
The work is a significant step towards establishing exponentially decaying diffuse distance fields as bridges between low level visual processing and shape computations.
Similar content being viewed by others
References
Ambrosio, L., Tortorelli, V.: On the approximation of functionals depending on jumps by elliptic functionals via Γ-convergence. Commun. Pure Appl. Math. 43(8), 999–1036 (1990)
Aslan, C., Tari, S.: An axis-based representation for recognition. In: ICCV, pp. 1339–1346 (2005)
Aslan, C., Erdem, A., Erdem, E., Tari, S.: Disconnected skeleton: Shape at its absolute scale. IEEE Trans. Pattern Anal. 30(12), 2188–2203 (2008)
Aubry, M., Schlickewei, U., Cremers, D.: The wave kernel signature: A quantum mechanical approach to shape analysis. In: ICCV—Workshop on Dynamic Shape Capture and Analysis (2011)
Bai, X., Wang, B., Yao, C., Liu, W., Tu, Z.: Co-transduction for shape retrieval. IEEE Trans. Image Process. 21(5), 2747–2757 (2012)
Bajaj, C.L., Pascucci, V., Schikore, D.R.: The contour spectrum. In: Proceedings of the 8th conference on Visualization (1997)
Ballester, C., Caselles, V., Igual, L., Garrido, L.: Level lines selection with variational models for segmentation and encoding. J. Math. Imaging Vis. 27(1), 5–27 (2007)
Bar, L., Sochen, N., Kiryati, N.: Image deblurring in the presence of impulsive noise. Int. J. Comput. Vis. 70(3), 279–298 (2006)
Biasotti, S., Cerri, A., Frosini, P., Giorgi, D., Landi, C.: Multidimensional size functions for shape comparison. J. Math. Imaging Vis. 32(2), 161–179 (2008)
Braides, A.: Approximation of Free-discontinuity Problems. Lecture Notes in Mathematics, vol. 1694. Springer, Berlin (1998)
Buades, A., Coll, B., Morel, J.M.: A non-local algorithm for image denoising. In: CVPR, pp. 60–65. Springer, Berlin (2005)
Burgeth, B., Weickert, J., Tari, S.: Minimally stochastic schemes for singular diffusion equations. In: Tai, X.C., Lie, K.A., Chan, T.F., Osher, S. (eds.) Image Processing Based on Partial Differential Equations, Mathematics and Visualization, pp. 325–339. Springer, Berlin (2006)
Chan, T., Vese, L.: Active contours without edges. IEEE Trans. Image Process. 10(2), 266–277 (2001)
Cremers, D., Tischhäuser, F., Weickert, J., Schnörr, C.: Diffusion snakes: Introducing statistical shape knowledge into the Mumford-Shah functional. Int. J. Comput. Vis. 50(3), 295–313 (2002)
Dimitrov, P., Lawlor, M., Zucker, S.: Distance images and intermediate-level vision. In: SSVM, pp. 653–664. Springer, Berlin (2011)
Droske, M., Rumpf, M.: Multi scale joint segmentation and registration of image morphology. IEEE Trans. Pattern Anal. 29(12), 2181–2194 (2007)
Edelsbrunner, H., Letscher, D., Zomorodian, A.: Topological persistence and simplification. Discrete Comput. Geom. 28, 511–533 (2002)
Erdem, E., Tari, S.: Mumford-Shah regularizer with contextual feedback. J. Math. Imaging Vis. 33(1), 67–84 (2009)
Erdem, E., Sancar-Yilmaz, A., Tari, S.: Mumford-Shah regularizer with spatial coherence. In: SSVM, pp. 545–555. Springer, Berlin (2007)
Gebal, K., Bærentzen, J.A., Aanæs, H., Larsen, R.: Shape analysis using the Auto Dinfusion Function. Comput. Graph. Forum 28, 1405–1413 (2009)
Gilboa, G., Darbon, J., Osher, S., Chan, T.: Nonlocal convex functionals for image regularization. UCLA CAM-report 06-57, (2006)
Gorelick, L., Galun, M., Sharon, E., Basri, R., Brandt, A.: Shape representation and classification using the Poisson equation. IEEE Trans. Pattern Anal. 28(12), 1991–2005 (2006)
Jin, Y., Jost, J., Wang, G.: A nonlocal version of the Osher-Sol-Vese model. J. Math. Imaging Vis. 44, 99–113 (2012)
Jung, M., Vese, L.: Nonlocal variational image deblurring models in the presence of Gaussian or impulse noise. In: SSVM, pp. 401–412. Springer, Berlin (2009)
Jung, M., Bresson, X., Chan, T., Vese, L.: Color image restoration using nonlocal Mumford-Shah regularizers. In: EMMCVPR, pp. 373–387. Springer, Berlin (2009)
Kontschieder, P., Donoser, M., Bischof, H.: Beyond pairwise shape similarity analysis. In: ACCV 2009. Lecture Notes in Computer Science, vol. 5996, pp. 655–666. Springer, Berlin (2010)
Lee, T.S., Yuille, A.: Efficient coding of visual scenes by grouping and segmentation. In: Doya, K., Ishii, S., Pouget, A., Rao, R. (eds.) Bayesian Brain: Probabilistic Approaches to Neural Coding, pp. 141–185. MIT Press, New York (2007)
Lee, T.S., Mumford, D., Romero, R., Lamme, V.A.: The role of the primary visual cortex in higher level vision. Vis. Res. 38(15–16), 2429–2454 (1998)
March, R., Dozio, M.: A variational method for the recovery of smooth boundaries. Image Vis. Comput. 15(9), 705–712 (1997)
Meyer, F.: Topographic distance and watershed lines. Signal Process. 38, 113–125 (1994)
Morse, S.P.: Concepts of use in contour map processing. Commun. ACM 12(3), 147–152 (1969)
Mumford, D., Shah, J.: Optimal approximations by piecewise smooth functions and associated variational problems. Commun. Pure Appl. Math. 42, 577–685 (1989)
Patz, T., Preusser, T.: Ambrosio-Tortorelli segmentation of stochastic images. In: ECCV, pp. 254–267. Springer, Berlin (2010)
Patz, T., Kirby, R., Preusser, T.: Ambrosio-Tortorelli segmentation of stochastic images: model extensions, theoretical investigations and numerical methods. Int. J. Comput. Vis. (2012). doi:10.1007/s11263-012-0578-8, 23 pp.
Pelillo, M., Siddiqi, K., Zucker, S.: Matching hierarchical structures using association graphs. IEEE Trans. Pattern Anal. 21(11), 1105–1120 (1999)
Peng, T., Jermyn, I., Prinet, V., Zerubia, J.: Extended phase field higher-order active contour models for networks. Int. J. Comput. Vis. 88(1), 111–128 (2010)
Pien, H., Desai, M., Shah, J.: Segmentation of MR images using curve evolution and prior information. Int. J. Pattern Recognit. 11(8), 1233–1245 (1997)
Preußer, T., Droske, M., Garbe, C., Rumpf, M., Telea, A.: A phase field method for joint denoising, edge detection and motion estimation. SIAM J. Appl. Math. 68(3), 599–618 (2007)
Proesman, M., Pauwels, E., van Gool, L.: Coupled geometry-driven diffusion equations for low-level vision. In: Romeny, B. (ed.) Geometry Driven Diffusion in Computer Vision. Lecture Notes in Computer Science. Kluwer, Amsterdam (1994)
Reuter, M.: Hierarchical shape segmentation and registration via topological features of Laplace-Beltrami eigenfunctions. Int. J. Comput. Vis. 89(2), 287–308 (2010)
Rosin, P.L., West, G.: Salience distance transforms. Graph. Models Image Process. 57(6), 483–521 (1995)
Rosman, G., Bronstein, M.M., Bronstein, A.M., Kimmel, R.: Nonlinear dimensionality reduction by topologically constrained isometric embedding. Int. J. Comput. Vis. 89(1), 56–68 (2010)
Shah, J.: Segmentation by nonlinear diffusion. In: CVPR, pp. 202–207. Springer, Berlin (1991)
Shah, J.: A common framework for curve evolution, segmentation and anisotropic diffusion. In: CVPR, pp. 136–142. Springer, Berlin (1996)
Shah, J.: Skeletons and segmentation of shapes. Tech. rep, Northeastern University (2005). See http://www.math.neu.edu/~shah/publications.html
Shah, J., Pien, H., Gauch, J.: Recovery of shapes of surfaces with discontinuities by fusion of shading and range data within a variational framework. IEEE Trans. Image Process. 5(8), 1243–1251 (1996)
Sun, J., Ovsjanikov, M., Guibas, L.: A concise and provably informative multi-scale signaturebased on heat diffusion. In: Comput. Graph. Forum (2009)
Tari, S.: Hierarchical shape decomposition via level sets. In: ISMM, pp. 215–225. Springer, Berlin (2009)
Tari, S.: Fluctuating distance fields. In: Breuss, M., Bruckestein, A., Maragos, P. (eds.) Innovations in Shape Analysis—Proceedings of Dagstuhl Workshop, Mathematics and Visualization. Springer, Berlin (2013)
Tari, S., Genctav, M.: From a modified Ambrosio-Tortorelli to a randomized part hierarchy tree. In: SSVM, pp. 267–278. Springer, Berlin (2011)
Tari, S., Shah, J.: Local symmetries of shapes in arbitrary dimension. In: ICCV, pp. 1123–1128 (1998)
Tari, S., Shah, J., Pien, H.: Extraction of shape skeletons from grayscale images. Comput. Vis. Image Underst. 66(2), 133–146 (1997)
Teboul, S., Blanc-Fraud, L., Aubert, G., Barlaud, M.: Variational approach for edge preserving regularization using coupled PDE’s. IEEE Trans. Image Process. 7, 387–397 (1998)
Yang, X., Bai, X., Koknar-Tezel, S., Latecki, J.: Densifying distance spaces for shape and image retrieval. J. Math. Imaging Vis. (2012). doi:10.1007/s10851-012-0363-x
Zhu, S.C., Yuille, A.L.: FORMS: a flexible object recognition and modeling system. Int. J. Comput. Vis. 20(3), 187–212 (1996)
Zucker, S.: Distance images and the enclosure field: applications in intermediate-level computer and biological vision. In: Breuss, M., Bruckestein, A., Maragos, P. (eds.) Innovations in Shape Analysis—Proceedings of Dagstuhl Workshop, Mathematics and Visualization. Springer, Berlin (2013)
Acknowledgements
This work has been partially funded by TUBITAK grant 112E208, the Alexander von Humboldt Foundation, and TUBITAK-BIDEB fellowship.
Author information
Authors and Affiliations
Corresponding author
Additional information
Preliminary conference version introducing randomized part hierarchy tree has appeared in SSVM 2011. The non-local field has first presented in [48].
Appendix
Appendix
Field Computation
To keep implementation simple we combine (9a) and (9b), and multiply the inhomogeneity function f by ρ 2 as scaling affects neither the geometrical nor the topological features of the level curves. In discrete setting this gives
where \(\mathbb{L}_{*}\) denotes the discrete Laplace operator; ω i,j ≈ω(x=i⋅h x ,y=j⋅h y ) with h x and h y are spatial discretization step sizes that are taken as the pixel width. Based on our discussions in Sect. 2, we set ρ 2=|Ω|. Next, we define a relaxed scheme:
where A is the left hand side of (10) and τ is the relaxation parameter selected smaller than \(\frac{|\varOmega|}{ 4|\varOmega|+2} \). When implemented in parallel the ω i,j values required for A need to be taken from the n th step; if, however, the values are being updated sequentially, updated values are used for faster convergence.
Saddle Point Detection
For locating saddle points, we do not rely on indefiniteness of the Hessian. Instead, we find watershed regions by calling Matlab’s watershed routine, which uses Meyer’s method [30]. We eliminate all those watershed boundaries that do not neighbor Ω +. On each of the remaining watershed boundaries, the saddle point is the minimum of the restriction of ω to the respective watershed boundary. Typically, the considered watershed boundaries extend from Ω + to the shape boundary. It may be possible, however, that a watershed boundary touching Ω + bifurcates before reaching the shape boundary. In this case there are indeed two watershed boundaries; the respective saddle points are given by the respective minima after the bifurcation.
Tree Matching
Let (V i ,E i ), i=1,2 be two rooted trees. Let k,l∈V 1 and m,n∈V 2 be distinct nodes of the respective trees. The tree association graph of the two trees is the graph (V,E) where V:=V 1×V 2, and the graph nodes (k,m)∈V and (l,n)∈V are adjacent when the connectivity between k and l is equivalent to that of m and n. Specifically, we say (k,m)∈V and (l,n)∈V are adjacent if level(k)−level(l)=level(m)−level(n) and the length of the path from k to l in the first tree is the same with the length of the path from m to n in the second tree.
Defining the equivalence between two sets of nodes in respective trees by comparing levels and path lengths, there exists a bijection between maximal subtree isomorphism and maximal clique of the association graph of the two trees; i.e., tree matching is equivalent to finding the maximal clique in the association graph. If the trees are attributed—e.g. in our case (V i ,E i ,α) where α is a function that assigns an attribute vector [α (1)(u) , α (2)(u)]T to each node u in either tree—then subtree isomorphism with the largest similarity is called maximum similarity subtree isomorphism. In this case, the weighted association graph is the weighted graph (V,E,c) such that c(z) for z≡(u,v), z∈V, u∈V 1 and v∈V 2 is defined via a similarity measure \(\operatorname{sim}(\cdot,\cdot)\) in the attribute space: \(c(z) = \operatorname{sim} (\alpha(u), \alpha(v))\). The attributes and similarity measure are calculated as described in Sect. 4.1.
Suitably defining a weight matrix M using node weights c(⋅), the global maximizer of x T Mx gives the maximum weight clique, which is solved iteratively:
where n is the iteration variable. The matrix M=(m ij ) is given via a matrix B=(b ij ) as follows:
where
Let maximum weighted clique be C⊂V. The solution x ∗ to the maximization problem (via the iterative scheme (11)) is expected to be
The iterative scheme (11) returns an approximation to the limit vector x ∗ in (12). What remains is how to interpret this vector. The paper [35] provides no suggestion on this. We adopt the following strategy instead of simply thresholding.
We start with an empty clique. Then starting with the node with the highest x value, we gradually add nodes to the clique in the order of decreasing x value. After each inclusion we compute expected vector using (12) and check the difference between this estimate and the actual vector returned by the iterative scheme (11). We keep adding nodes till inclusion of nodes no longer decreases the difference.
Rights and permissions
About this article
Cite this article
Tari, S., Genctav, M. From a Non-Local Ambrosio-Tortorelli Phase Field to a Randomized Part Hierarchy Tree. J Math Imaging Vis 49, 69–86 (2014). https://doi.org/10.1007/s10851-013-0441-8
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10851-013-0441-8