fjdhkyf3423/AdaptiveRateCompressiveForegroundExtraction

Code to replicate experiments in paper

Adaptive Rate Reconstruction of Time Varying Signals With Applications in Compressive Foreground Extraction

This code replicates the experiments in

1. Adaptive-Rate Reconstruction of Time-Varying Signals with Application in Compressive Foreground Extraction.
   J. F. C. Mota, N. Deligiannis, A. C. Sankaranarayanan, V. Cevher, M. R. D. Rodrigues.
   IEEE Transactions on Signal Processing, Vol. 64, No. 14, pp. 3651-3666, 2016.
   link, arXiv

and

2. Dynamic Sparse State Estimation Using L1-L1 Minimization: Adaptive-Rate Measurement Bounds, Algorithms and Applications.
   J. F. C. Mota, N. Deligiannis, A. C. Sankaranarayanan, V. Cevher, M. R. D. Rodrigues.
   IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Brisbane, 2015.
   link

Organization

• createFigures: Replicates the figures in [1]. As all figures were postprocessed in LaTeX, the visuals are not exactly as in the paper. Some experiments only generate data, not figures.

  • FFTMeasurements: Generates data to create Figure 4 in [1].

  • GaussianMeasurements: Generates data to create Figures 3(a), 3(b), 3(e), and 3(f) in [1].

  • GaussianMeasurementsNoise: Generates data to create Figures 3(c) and 3(d).

• Algorithms: Code to solve the L1-L1 minimization problem:

  

  There are two implementations:

  • basisPursuitPlusL1/ADMM:
    Solver used in the experiments in

    Compressed Sensing with Prior Information: Strategies, Geometry, and Bounds.
    J. F. C. Mota, N. Deligiannis, M. R. D. Rodrigues.
    IEEE Transactions on Information Theory, Vol. 63, No. 7, pp. 4472-4496, 2017.
    link, arXiv, Github page

  • PrimalDualFramework-TranDinhCevher/DECOPT:
    Solver proposed in

    Constrained convex minimization via model-based excessive gap.
    Q. Tran-Dinh, V. Cevher.
    Proceedings of the annual conference on Neural Information Processing Systems Foundation (NIPS), 2014.
    link, see also code webpage

    The original code was slightly modified to be applicable to L1-L1 minimization.

  • otherApproaches/modifiedCS-Vaswani/Modified-CS:
    Solver for the Modified-CS problem:

    

    which was proposed in

    Modified-CS: Modifying Compressive Sensing for Problems With Partially Known Support.
    N. Vaswani, W. Lu.
    IEEE Transactions on Signal Processing, Vol. 58, No. 9, 2010.
    link

  • datasets: Contains .mat files with the video sequences used in [1]. For further information about the datasets check the README.TXT file in that folder.

---

License: GPLv3