A dynamic generating graphical model for point-sets matching
Introduction
Point-sets matching is a fundamental method for graph matching, which is a task of pivotal importance in high level computer vision with many applications such as image retrieval [1], [2], object tracking [3], panoramic mosaic [4], [5], object recognition [6], [7], computational biology and computational chemistry [8], [9].
Point-sets matching consists of finding correspondences between two sets of interest points, which are extracted from two partially overlapping images taken from different views. In this paper, the problem of point-sets matching is defined as follows. Assume we have two images to be matched, D and T, which are defined as the data image and target image. The set of interest points derived from the data image is denoted as and that from the target image as , with ND and NT as the number of points in each set. The goal is to find the best mapping between a subset of S1 and a subset of S2. Here we follow the principle that it is better to have a small number of matches than to have wrong matches [10].
Many researchers have attempted to obtain more accurate results and reduce the complexity of the computation. In the computer vision literature there have been several categories of solutions including spectral methods [11], [12], [13], tree search [14], [15], [16], [17], relaxation labeling [18], [19], [20], [21] and various others [22], [23], [24]. Based on the spectral graph theory [25], [26], the main idea behind spectral methods is that the eigenvalues and the eigenvectors of the adjacency matrix of a graph are invariant with respect to node permutations [27]. An important limitation of spectral methods is that when there are spurious nodes and edges in the graphs under study, the performance will soon break down. Tree search methods generate a matching tree by iteratively adding new pairs of corresponding nodes from the initial tree (usually empty) and make comparison after each addition of nodes to eliminate incorrect assignment. The number of possible assignments is very large if there are many nodes to be matched. Relaxation labeling algorithms assign each node in the data image a label of probability, which can indicate which node in the target image is the best matching result for it. The initial probability is computed by local information of each node, and then it is iteratively updated taking into account the neighboring nodes until a global maximum probability of each node is reached. The final probability will determine for each node in the data image which node of the other image it corresponds to. Relaxation labeling methods have a much lower computational complexity than the tree search method.
Our work is based on the theory of probabilistic graphical models. Inference in graphical models is a method of global optimality and is optimal in a maximum a posteriori sense, but is limited in application due to its NP-complete [28] complexity. Recently, Caetano et al. [29] presented a probabilistic graphical model for point-sets matching, which was proven to be optimal in the maximum a posteriori sense and has polynomial time dependency on the point set sizes. Comparison [30] has been made between this method and standard probabilistic relaxation labeling (PRL) [21] using different forms of point metrics and under different levels of additive noise. The comparison showed that Caetano’s method is more effective than PRL.
The constraint to Caetano’s model is that the mapping must be a total function: every point in the data image must be mapped to one point in the target image, that is, no outliers are allowed. But in many applications such as panoramic mosaics, this constraint may not be applicable; for example, there may be more or fewer outliers existing in the real world. The outliers are the feature points in either image that have no counterparts in the other one, while “signal” points in this paper are defined as feature points which are not outliers. Fig. 1 shows two images to be mapped (Fig. 1a is defined as a data image, while Fig. 1b is defined as a target image), with feature points detected by the Harris corner detector. In these two images, “signal” points are denoted by ∗, and outliers are denoted by □. Theoretically, the point matching between the two images can not be solved using Caetano’s model. The experimental results in Section 4 show that the performance of Caetano’s algorithm worsens when there are more outliers increase in the data image.
In this paper, we discuss the reason that Caetano’s algorithm soon breaks down when there are outliers in the data image, and draw the conclusion that the selection of the separators in the graphical models will greatly influence the result of the mapping. We propose a new algorithm is brought forward to generate graphical models that can ensure an optimal selection of separators. The results demonstrate that the proposed technique outperforms Caetano’s model.
Section snippets
Theory of graphical models and Junction Tree
A graphical model (Fig. 2a) is a graph endowed with joint probability distribution. The nodes in a graphical model represent random variables while the edges represent dependence between the nodes.
A clique (denoted by ellipses in Fig. 2b) is a maximal subgraph with every pair of vertices joined. A Junction Tree (Fig. 2b) is a graph where the nodes correspond to the cliques such that the running intersection property is satisfied. This property states that if a variable is contained in two
The separators’ influence in Junction Tree inference
The Hugin algorithm works in two steps: initialization and message-passing. In the first step, the “clique potentials” are generated. In the second step, information is transferred between each clique. For all the cliques to be consistent with each other, we only need to ensure local consistency between neighboring cliques.
Suppose that we have two adjacent cliques A (consisting of nodes x1, x2, x3, and x4) and B (consisting of nodes x1, x2, x3, and x5) in a Junction Tree through a separator set
Experiments and result
We carried out three experiments. In the first experiment, we generated both data and target images, 256 × 256 pixels. Ten pairs of “signal” points were added to both the images. The noise in each pair of “signal” points was created by adding independent random numbers drawn from a normal distribution with zero mean and standard deviation 2 to both x and y coordinates. Then, we added 10 outliers to the target image and various numbers of outliers, from 1 to 10, to the data image. The 10 pairs of
Conclusion
In this paper, we have discussed how the separators influence message-passing in the Junction Tree inference. Two ways were considered to improve the existing algorithm to select optimal separators. The combination of an initializing algorithm and an updating algorithm provides performance that is the least influenced by the outliers in the data image. We presented a new process in which bi-mapping is conducted after forward mapping in order to solve the problem that occurs in outliers when
Future work
Our method has some shortcomings, which can be overcome in future work.
- 1.
The computing complexity of our algorithm is greatly reduced compared with fully connected graphical models, but it rapidly becomes more complex with the increasing size of point sets. For feature point sets larger than 50 points in each image, it is hard to assure real time matching. We will try layered matching for large point sets. In this method the feature points are separated into several small subsets.
- 2.
Currently our
References (32)
- et al.
A survey of medical image registration
IEEE Eng. Med. Biol. Mag.
(1998) - et al.
Matching point features under small nonrigid motion
Pattern Recogn.
(2001) - et al.
Spectral correspondence for point pattern matching
Pattern Recogn.
(2003) - et al.
A graph distance metric combining maximum common sub graph and minimum common supergraph
Pattern Recogn. Lett.
(2001) Structural pattern recognition using genetic algorithm
Pattern Recogn.
(2002)- et al.
Point matching under nonuniform distortions
Discrete Appl. Math.
(2003) - et al.
Local gray value invariants for image retrieval
IEEE Trans. Pattern Anal. Mach. Intelligence
(1997) - V. Gouet, N. Boujemaa, On the robustness of color points of interest for image retrieval, in: Proceedings of...
Effective tracking through tree-search
IEEE Trans. Pattern Anal. Mach. Intelligence
(2003)- M. Brown, D.G. Lowe, Recognising panoramas, in: Proceedings of the Ninth IEEE International Conference on Computer...
Molecular surface recognition by a computer vision based technique
Protein Eng.
Structural graph matching using the EM algorithm and singular value decomposition
IEEE Trans. Pattern Anal. Mach. Intelligence
Correspondence matching with model clusters
IEEE Trans. Pattern Anal. Mach. Intelligence
Cited by (7)
A novel probabilistic model based fingerprint recognition algorithm
2012, Procedia EngineeringImproved point pattern matching method using dynamic generating graphical model
2013, Guangdian Gongcheng/Opto-Electronic EngineeringPoint pattern matching algorithm with O(n log n) time
2011, ICIC Express LettersA Fingerprint Matching Algorithm Based on Probabilistic Graphical Model and 3-Tree Model
2011, Proceedings of SPIE - The International Society for Optical EngineeringA novel fingerprint matching scheme combining graphical model and thin plate spline model
2009, Proceedings of SPIE - The International Society for Optical Engineering