Fast compressed sensing reconstruction using the least squares and signal correlation

2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

2014

Our recent work has shown that quality of compressed sensing reconstruction can be improved immensely by minimising the error between the signal and a correlated reference, as opposed to the conventional l 1 -minimisation of the data measurements. This paper introduces a method for online estimating suitable references for video sequences using the running Gaussian average. The proposed method can provide robustness to video content changes as well as reconstruction noise. The experimental results demonstrate the performance of this method to be superior to those of the state-of-the-art l 1 -min methods. The results are comparable to the lossless reference reconstruction approach.

“…The proof can be found in [14]. This means the closer the reference r is to the original x, the better reconstruction result can be obtained.…”

Section: Reconstruction Using Correlated Referencesmentioning

confidence: 88%

Section: Reconstruction Using Correlated Referencesmentioning

confidence: 99%

“…(3) can also be found in [14]. The use of the least squares method can provide a much lower complexity compares to the l 1 -min.…”

Section: Reconstruction Using Correlated Referencesmentioning

confidence: 99%

Section: Implementation Detailsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Running Gaussian reference-based reconstruction for video compressed sensing

2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

2014

An optimal learning parameter for running Gaussian-based referenced compressive

2nd IET International Conference on Intelligent Signal Processing 2015 (ISP)

2015

Self Cite

One of the approaches to exploit temporal redundancy in compressive sensing reconstruction of spatio-temporal signals is the Running Gaussian-based Referenced Compressive Sensing. It uses the weighted-average of all prior reconstructed instances as a reference to reconstruct the next instance with high accuracy. The performance of this approach depends on the weight called learning parameter. This work studies the relationship between the learning parameter and the reconstruction accuracy. We show that the small value of the learning parameter is more suitable for natural signals with dynamic sparse supports. We also propose a dynamic optimal learning parameter that provides good reconstruction accuracy for all signals. Out experimental results show that the proposed optimal learning parameter outperforms all fixed values of learning parameter in natural video sequences reconstruction.

Towards baseline-independent analysis of compressive sensed functional magnetic resonance image data

2017 22nd International Conference on Digital Signal Processing (DSP)

2017

The main task of Functional Magnetic Resonance Imaging (fMRI) is the localisation of brain activities, which depends on the detection of hemodynamic responses in the Blood Oxygenation-Level Dependent (BOLD) signal. While compressive sensing has been widely applied to improve the quality and resolution of MRI in general, its reconstruction noise overwhelms the small magnitude of hemodynamic responses. We propose a new reconstruction algorithm for the compressive sensing fMRI that exploits the temporal redundancy of the data, called Referenced Compressive Sensing, which works well in preserving fMRI analytical features. We also propose the use of the baselineindependent signal for analysis of reconstructed data. It is shown that the baseline-independent reconstructed data from Referenced Compressive Sensing is highly correlated to the lossless data, thus preserving more of the analytical features.