A comparison of six deconvolution techniques

Madden, Francis N.; Godfrey, K.R.; Chappell, Michael J.; Hovorka, Roman; Bates, R. A.

doi:10.1007/bf02353672

Cited by 63 publications

(53 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Regularization is another possibility to "force" the inverse filtered object function to have some prescribed properties. 33,34 It is worth mentioning a comparative study of six different deconvolution techniques by Madden et al, 35 where the reader can find a numerical analysis of some specific methods used to deconvolve pharmacokinetic drug response functions. However, these methods apply either an implicit model function (e.g., cubic splines) or an interpolation of the original experimental data to a substantially increased grid size compared to the number of measured points.…”

Section: Deconvolution Methodsmentioning

confidence: 99%

Model-Free Deconvolution of Femtosecond Kinetic Data

Bányász

Keszei

2006

J. Phys. Chem. A

View full text Add to dashboard Cite

Though shorter laser pulses can also be produced, pulses of the 100 fs range are typically used in femtosecond kinetic measurements, which are comparable to characteristic times of the studied processes, making detection of the kinetic response functions inevitably distorted by convolution with the pulses applied. A description of this convolution in terms of experiments and measurable signals is given, followed by a detailed discussion of a large number of available methods to solve the convolution equation to get the undistorted kinetic signal, without any presupposed kinetic or photophysical model of the underlying processes. A thorough numerical test of several deconvolution methods is described, and two iterative time-domain methods (Bayesian and Jansson deconvolution) along with two inverse filtering frequency-domain methods (adaptive Wiener filtering and regularization) are suggested to use for the deconvolution of experimental femtosecond kinetic data sets. Adaptation of these methods to typical kinetic curve shapes is described in detail. We find that the modelfree deconvolution gives satisfactory results compared to the classical "reconvolution" method where the knowledge of the kinetic and photophysical mechanism is necessary to perform the deconvolution. In addition, a model-free deconvolution followed by a statistical inference of the parameters of a model function gives less biased results for the relevant parameters of the model than simple reconvolution. We have also analyzed real-life experimental data and found that the model-free deconvolution methods can be successfully used to get undistorted kinetic curves in that case as well. A graphical computer program to perform deconvolution via inverse filtering and additional noise filters is also provided as Supporting Information. Though deconvolution methods described here were optimized for femtosecond kinetic measurements, they can be used for any kind of convolved data where measured experimental shapes are similar.

show abstract

Section: Deconvolution Methodsmentioning

confidence: 99%

Model-Free Deconvolution of Femtosecond Kinetic Data

Bányász

Keszei

2006

J. Phys. Chem. A

View full text Add to dashboard Cite

show abstract

“…Madden et al (1996) undertook a comparison of six possible deconvolution techniques for solving Equation 13. Of the six approaches, they concluded that the best in terms of overall performance was 'maximum entropy deconvolution', developed by Skilling and Bryan (1982).…”

Section: Deconvolutionmentioning

confidence: 99%

Residence Time Distributions for Turbulent, Critical, and Laminar Pipe Flow

et al. 2016

View full text Add to dashboard Cite

Longitudinal dispersion processes are often described by the Advection Dispersion Equation (ADE), which is analogous to Fick's law of diffusion, where the impulse response function of the spatial concentration distribution is assumed to be Gaussian. This paper assesses the validity of the assumption of a Gaussian impulse response function, using Residence Time Distributions (RTDs) obtained from new laboratory data. Measured up-and down-stream temporal concentration proÞles have been deconvolved to numerically infer RTDs for a range of turbulent, critical and laminar pipe ßows.It is shown that the Gaussian impulse response function provides a good estimate of the system's mixing characteristics for turbulent and critical ßows, and an empirical equation to estimate the dispersion coefficient for Reynolds Number, Re, between 3,000 and 20,000 is presented. For laminar ßow, here identified as Re < 3000, the RTDs do not conform to the Gaussian assumption due to insufficient time being available for the solute to become cross-sectionally well mixed. For this situation, which occurs commonly in water distribution networks, a theoretical RTD for laminar ßow that assumes no radial mixing is shown to provide a good approximation of the system's mixing characteristics at short times after injection.

show abstract

“…Compared to all these previous techniques, the entropy maximizations algorithm [14,15] performs better in reducing the RMS error on the solution [16].…”

Section: Introductionmentioning

confidence: 97%

Diffraction effects detection for HDR image-based measurements

Lucat¹,

Hegedüs²,

Pacanowski³

2017

Opt. Express

View full text Add to dashboard Cite

Modern imaging techniques have proved to be very efficient to recover a scene with high dynamic range (HDR) values. However, this high dynamic range can introduce star-burst patterns around highlights arising from the diffraction of the camera aperture. The spatial extent of this effect can be very wide and alters pixels values, which, in a measurement context, are not reliable anymore. To address this problem, we introduce a novel algorithm that, utilizing a closed-form PSF, predicts where the diffraction will affect the pixels of an HDR image, making it possible to discard them from the measurement. Our approach gives better results than common deconvolution techniques and the uncertainty values (convolution kernel and noise) of the algorithm output are recovered. References and links1. E. Reinhard, High dynamic range imaging : acquisition, display, and image-based lighting (Morgan Kaufmann/Elsevier, 2010). 2. A. N. A. N. Tikhonov and V.

show abstract

A comparison of six deconvolution techniques

Cited by 63 publications

References 17 publications

Model-Free Deconvolution of Femtosecond Kinetic Data

Model-Free Deconvolution of Femtosecond Kinetic Data

Residence Time Distributions for Turbulent, Critical, and Laminar Pipe Flow

Diffraction effects detection for HDR image-based measurements

Contact Info

Product

Resources

About