Multiscale Matching Research Articles

Multiscale contour matching with basis functions

A multiscale contour matching method using basis functions is presented. In this scheme, curves are parametrized by the curve length s and s′, and contours are matched by finding a matching function s′=f(s). A set of basis functions, which are monotonic and continuous in the interval [0,1], are designed. The sum of these weighted basis functions approximates f(s). The weight of each function can be computed through optimization of the objective function. Multiscale schemes can make matching insensitive to initial value. Experimental validation is carried out with a sequence of human active contours and the result shows this method is effective.

Measurement and Management for Medical and Clinical Risks

Hosptial information system (HIS) collects all the data from all the branches of departments in a hospital, including laboratory tests, physiological tests, electronic patient records. Thus, HIS can be viewed as a large heterogenous database, which stores chronological changes in patients' status. This paper overviews three applications of data mining and statistical methods to HIS. First, clustering of temporal sequences based on multiscale matching was applied for grouping chronic hepatitis. Second, decision tree method was used for detection of risk factors, which was successfully used to prevent nursing medication errors. Finally, several linear models were applied for hospital management data. These results show that data mining methods, including decision tree mining, temporal data mining, are useful for detection of risk factors from large distributed data such as HIS, whose process can be called risk mining.

Quaternion wavelet phase based stereo matching for uncalibrated images

This paper proposes a quaternion wavelet phase based stereo matching (QWPSM) scheme for uncalibrated image pairs. In this scheme, we estimate the disparity by directly establishing correspondences between quaternionic phase structures of two quaternion wavelet filtered (QWF) images. Firstly, linear-phase quaternion wavelet filters (LPQWFs) are constructed from real biorthogonal wavelet bases. Then, quaternion phases are extracted under each scale through quaternion wavelet filtering of the multiscale transformed image pyramids. The disparity estimation is formed as a minimization process of a local energy weighted cost function, and propagated from coarse to fine scales. Costs can adaptively alleviate the negative effects of phase singularities, which are the main causes of mismatches in phase-based stereo matching. Multiscale matching strategy is used to avoid phase wrapping and improve convergence speed. Experimental results are promising in various image pairs.

Automated discovery of chronological patterns in long time-series medical datasets

Data mining in time-series medical databases has been receiving considerable attention because it provides a way of revealing useful information hidden in the database, for example, relationships between the temporal course of examination results and the onset time of diseases. This article presents a new method for finding similar patterns in temporal sequences. The method is a hybridization of phase-constraint multiscale matching and rough clustering. Multiscale matching enables us to cross-scale a comparison of the sequences, namely, it enables us to compare temporal patterns by partially changing observation scales. Rough clustering enables us to construct interpretable clusters of the sequences even if their similarities are given as relative similarities. We combine these methods and cluster the sequences according to the multiscale similarity of patterns. Experimental results on the chronic hepatitis dataset showed that clusters demonstrating interesting temporal patterns were successfully discovered. © 2005 Wiley Periodicals, Inc. Int J Int Syst 20: 737–757, 2005.

A multiresolution approach to orientation assignment in 3D electron microscopy of single particles

Three-dimensional (3D) electron microscopy (3DEM) aims at the determination of the spatial distribution of the Coulomb potential of macromolecular complexes. The 3D reconstruction of a macromolecule using single-particle techniques involves thousands of 2D projections. One of the key parameters required to perform such a 3D reconstruction is the orientation of each projection image as well as its in-plane orientation. This information is unknown experimentally and must be determined using image-processing techniques. We propose the use of wavelets to match the experimental projections with those obtained from a reference 3D model. The wavelet decomposition of the projection images provides a framework for a multiscale matching algorithm in which speed and robustness to noise are gained. Furthermore, this multiresolution approach is combined with a novel orientation selection strategy. Results obtained from computer simulations as well as experimental data encourage the use of this approach.

Multidimensional signal compression using multiscale recurrent patterns

In this paper we propose a new multidimensional signal lossy compression method based on multiscale recurrent patterns, referred to as multidimensional multiscale parser (MMP). In it, a multidimensional signal is recursively segmented into variable-length vectors, and each segment is encoded using expansions and contractions of vectors in a dictionary. The dictionary is updated while the data is being encoded, using concatenations of expanded and contracted versions of previously encoded vectors. The only data encoded are the segmentation tree and the indexes of the vectors in the dictionary, and therefore no side information is necessary for the dictionary updating. The signal segmentation is carried out through a rate–distortion optimization procedure. A two-dimensional version of the MMP algorithm was implemented and tested with several kinds of image data. We have observed that the proposed dictionary updating procedure is effective in adapting the algorithm to a large variety of image content, lending to it a universal flavor. For text and graphics images, it outperforms the state-of-the-art SPIHT algorithm by more that 3 dB at 0.5 opp , while for mixed document images, containing text, graphics and gray-scale images, by more than 1.5 dB at the same rate. Due to the way the images are segmented, they appear slightly blocky at low rates. We have alleviated this problem by proposing an effective way of reducing the blockiness in the reconstructed image, with no penalty in signal-to-noise ratio performance in most cases. We conclude the paper with a theoretical analysis of the approximate matching of Gaussian vectors using scales, which gives a justification of why approximate multiscale matching is a good option, specially at low rates.

Quantizacao Vetorial Adaptativa Multiescalas com Otimização Taxa-Distorção

In this paper we propose a new multidimensional lossy sigual compression method based on multiscale recurrent patterns, referred to as MMP (Multidimensional Multiscale Parser). In it, a multidimensional signal is recursively segmented into variable-length vectors, and each segment is encoded using expansions and contractions of vectors in a dictionary. The dictionary is updated while the data is being encoded, using concatenations of expanded and contracted versions of previously encoded vectors. The only data encoded are the segmentation tree and the indexes of the vectors in the dictionary, and therefore no side information is necessary for the dictionary updating. The signal segmentation is carried out through a rate-distortion optimization procedure. A two-dimensional version of the MMP algorithm was implemented and tested with several kinds of image data. We have observed that the proposed dictionary updating procedure is effective in adapting the algorithm to a large variety of image content, lending to it a universal flavor. For text and graphics images, it outperforms the state-of-the-art SPIHT algorithm by more that 3dB at 0.5bpp, while for mixed document images, containing text, graphics and gray scale images, by more than 1.5dB at the same rate. We conclude the paper with a theoretical analysis of the approximate matching of gaussian vectors using scales, which gives a justification of why approximate multiscale matching is a good option, speciallyat low rates.

Universal Multiscale Matching Pursuits

In this paper we describe the UMMP (Universal Multiscale Matching Pursuits), a practical universal algorithm for multi-dimensional data lossy compression. The method is based on approximate multiscale matching of recurrent patterns. It uses a dictionary of basis functions in which the data is decomposed in the spirit of Mallat's Matching Pursuits (MP). Unlike MP however, UMMP builds its own dictionary while encoding the data, instead of using a previously defined one. Also, all basis functions can be contracted or dilated to better match the input data during the expansion. This allows the algorithm to perform arbitrarily close to the source's D(R) function when the rate goes to zero. Simulation results show that it has good coding performance for a large class of data.