Abstract: Reconstruction method for reconstructing a first signal (x(t)) regularly sampled at a sub-Nyquist rate, comprising the step of retrieving from the regularly spaced sampled values (ys[n], y(nT)) a set of weights (cn, cnr, ck) and shifts (tn, tk) with which said first signal (x(t)) can be reconstructed. The reconstructed signal (x(t)) can be represented as a sequence of known functions (?(t)) weighted by the weights (ck) and shifted by the shifts (tk). The sampling rate is at least equal to the rate of innovation (?) of the first signal (x(t)).
Type:
Grant
Filed:
June 14, 2010
Date of Patent:
October 4, 2011
Assignee:
QUALCOMM Incorporated
Inventors:
Martin Vetterli, Pina Marziliano, Thierry Blu
Abstract: This disclosure is directed to techniques for memory efficient variable to fixed length (VF) coding techniques for a variety of applications, such as media coding. For example, such techniques may be applied to code digital video, image, audio, or speech data. The techniques described in this disclosure may utilize algebraic properties of VF coding trees constructed using Tunstall or Khodak algorithms, and employ combinatorial enumeration techniques for construction, encoding and decoding of codewords. For some applications, the VF coding techniques may be implemented within media encoders, decoders, or combined encoder-decoders (CODECs). Also, in some aspects, various attributes defining a VF coding tree constructed according to this disclosure may be provided in a memory efficient data structure stored in memory associated with a coding device.