Convolution on the -Sphere With Application to PDF Modeling

Dept. of Electron. Syst. & Inf. Process., Univ. of Zagreb, Zagreb, Croatia
IEEE Transactions on Signal Processing (Impact Factor: 3.2). 04/2010; DOI: 10.1109/TSP.2009.2033329
Source: IEEE Xplore

ABSTRACT In this paper, we derive an explicit form of the convolution theorem for functions on an n -sphere. Our motivation comes from the design of a probability density estimator for n -dimensional random vectors. We propose a probability density function (pdf) estimation method that uses the derived convolution result on Sn. Random samples are mapped onto the n -sphere and estimation is performed in the new domain by convolving the samples with the smoothing kernel density. The convolution is carried out in the spectral domain. Samples are mapped between the n -sphere and the n -dimensional Euclidean space by the generalized stereographic projection. We apply the proposed model to several synthetic and real-world data sets and discuss the results.

  • [Show abstract] [Hide abstract]
    ABSTRACT: We develop a new type of convolution between two signals on the 2-sphere. This is the first type of convolution on the 2-sphere which is commutative. Two other advantages, in comparison with existing definitions in the literature, are that 1) the new convolution admits anisotropic filters and signals and 2) the domain of the output remains on the sphere. Therefore, the new convolution well emulates the conventional Euclidean convolution. In addition to providing the new definition of convolution and discussing its properties, we provide the spectral analysis of the convolution output. This convolutional framework can be useful in filtering applications for signals defined on the 2-sphere.
    IEEE Transactions on Signal Processing 12/2012; 60(12):6697-6703. DOI:10.1109/TSP.2012.2217337 · 3.20 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: A new recursive fast radix-2 algorithm for an efficient computation of the modulated complex lapped transform (MCLT) is presented. Based on a new proposed alternative recursive sparse matrix factorization for the MDCT (modified discrete cosine transform) matrix and a relation between the MDCT and the MDST (modified discrete sine transform), firstly a new recursive fast radix-2 MDST algorithm is derived. The corresponding fast MDCT and MDST computational structures are regular and complementary to each other. Consequently, this fact enables us by their composition to construct a fast MCLT computational structure representing the fast recursive radix-2 MCLT algorithm. The fast MCLT computational structure is regular and all its stages may be realized in parallel. Combining the proposed fast radix-2 MCLT algorithm with an existing generalized fast mixed-radix MDCT algorithm defined for the composite lengths N = 2 × qm, m ≥ 2, where q is an odd positive integer, we can compute the MCLT for the composite lengths N = 2n × qm, n, m ≥ 2, thus supporting a wider range of transform sizes compared to existing fast MCLT algorithms.
    IEEE Transactions on Signal Processing 12/2012; 60(12):6703-6708. DOI:10.1109/TSP.2012.2213082 · 3.20 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper examines filtering on a sphere, by first examining the roles of spherical harmonic magnitude and phase. We show that phase is more important than magnitude in determining the structure of a spherical function. We examine the properties of linear phase shifts in the spherical harmonic domain, which suggest a mechanism for constructing finite-impulse-response (FIR) filters. We show that those filters have desirable properties, such as being associative, mapping spherical functions to spherical functions, allowing directional filtering, and being defined by relatively simple equations. We provide examples of the filters for both spherical and manifold data.
    IEEE Transactions on Signal Processing 08/2012; 60(12). DOI:10.1109/TSP.2012.2213083 · 3.20 Impact Factor