Search results for "Convolution"
showing 10 items of 334 documents
Deep Learning Techniques for Depression Assessment
2018
Depression is a typical mood disorder, which affects a significant number of individuals worldwide at an increasing rate. Objective measures for early detection of signs related to depression could be beneficial for clinicians with regards to a decision support system. In this paper, assessment of depression is done by applying three deep learning techniques of Convolutional Neural Network (CNN). These techniques are transfer learning using AlexNet, fine-tuning using AlexNet and building an end to end CNN. The inputs of the CNNs are a combination of Motion History Image, Landmark Motion History Image and Gabor Motion History Image, and have been generated on a depression dataset. Accuracy o…
Optimal extension of multispectral image demosaicking algorithms for setting up a one-shot camera video acquisition system
2022
Multispectral images are acquired using multispectral cameras equipped with CCD or CMOS sensors which sample the visible or near infrared spectrum according to specific spectral bands. A mosaic of multispectral MSFA filters is superimposed on the surface of the sensors to acquire a raw image called an MSFA image. In the MSFA image, only one spectral band is available per pixel, the demosaicking process is necessary to estimate the multispectral image at full spatio-spectral resolution. Motivated by the success of single-sensor cameras capturing the image in a single exposure that use CFA filters, we performed a comparative study of a few recent color image demosaicking algorithms and experi…
Numerical analysis of density gradient centrifugation profiles from eukaryotic DNA
1990
A numerical method for the deconvolution of superimposed Gaussian distributions with a unique solution has been proposed by Medgyessy [10]. We have tested the usefulness of this method for the analysis of density gradient centrifugation profiles from eukaryotic DNA, which are normally composed from overlapping Gaussian distributed profiles of several subcomponents with different mean buoyant densities. From the analysis of human DNA and from model calculations we conclude that major subcomponents can be identified by this method, if they differ in their buoyant density by approximatly 0.005 g/ml. Minor components can only be identified if the total DNA has been fractionated according to buo…
Focal plane array infrared camera transfer function calculation and image restoration
2004
Infrared images often present distortions induced by the measurement system. Image processing is thus an essential part of infrared measurements. A distortion model based on a convolution product is presented. The analytical form of the convolution kernel has been obtained from an image formation theory, along with an analysis of the sampling of the focal plane array camera detector's matrix. Image restoration is an ill-posed problem, and its solution can be obtained using regularization methods. In this work, image restoration is performed using a variation of Tikhonov regularization that makes use of the particular form of the convolution kernel matrix, which is built as a block-circulant…
The effects of convolution and gradient dependence on a parametric Dirichlet problem
2020
Our objective is to study a new type of Dirichlet boundary value problem consisting of a system of equations with parameters, where the reaction terms depend on both the solution and its gradient (i.e., they are convection terms) and incorporate the effects of convolutions. We present results on existence, uniqueness and dependence of solutions with respect to the parameters involving convolutions.
Using Wave Propagation Simulations and Convolutional Neural Networks to Retrieve Thin Film Thickness from Hyperspectral Images
2021
Ill-posed inversion problems are one of the major challenges when there is a need to combine measurements with the theory and numerical model. In this study, we demonstrate the use of wave propagation simulations to train a convolutional neural network (CNN) for retrieving sub-wavelength thickness profiles of thin film coatings from hyperspectral images. The simulations are produced by solving numerically one-dimensional wave equation with a method based on Discrete Exterior Calculus (DEC). This approach provides a powerful tool to produce large sets of training data for the neural network. CNN was verified by simulated verification sets and measured reflectance spectra, both of which showe…
On the symbol homomorphism of a certain Frechet algebra of singular integral operators
1985
We prove the surjectivity of the symbol map of the Frechet algebra obtained by completing an algebra of convolution and multiplication operators in the topology generated by all L2-Sobolev norms. The proof is based on an ℝn of Egorov's theorem valid for non-homogeneous principal symbols, discussed in [5], [6]. We use the hyperbolic equation ∂u/∂t=i|D|ηu, 0<η<1, which has its characteristic flow constant at infinity, so that no differentiability of the symbol is required there.
Functional calculi for convolution operators on a discrete, periodic, solvable group
2009
Suppose T is a bounded self-adjoint operator on the Hilbert space L2(X,μ) and let T=∫SpL2TλdE(λ) be its spectral resolution. Let F be a Borel bounded function on [−a,a], SpL2T⊂[−a,a]. We say that F is a spectral Lp-multiplier for T, if F(T)=∫SpL2TF(λ)dE(λ) is a bounded operator on Lp(X,μ). The paper deals with l1-multipliers, where X=G is a discrete (countable) solvable group with ∀x∈G, x4=1, μ is the counting measure and TΦ:l2(G)∋ξ↦ξ∗Φ∈l2(G), where Φ=Φ∗ is a l1(G) function, suppΦ generates G. The main result of the paper states that there exists a Ψ on G such that all l1-multipliers for TΨ are real analytic at every interior point of Spl2(G)TΨ. We also exhibit self-adjoint Φ′s in l1(G) suc…
Perturbations of surjective convolution operators
2002
Let μ 1 and μ 2 be (ultra)distributions with compact support which have disjoint singular supports. We assume that the convolution operator f → μ 1 *f is surjective when it acts on a space of functions or (ultra)distributions, and we investigate whether the perturbed convolution operator f→ (μ 1 + μ 2 ) * f is surjective. In particular we solve in the negative a question asked by Abramczuk in 1984.
The convolution operation on the spectra of algebras of symmetric analytic functions
2012
Abstract We show that the spectrum of the algebra of bounded symmetric analytic functions on l p , 1 ≤ p + ∞ with the symmetric convolution operation is a commutative semigroup with the cancellation law for which we discuss the existence of inverses. For p = 1 , a representation of the spectrum in terms of entire functions of exponential type is obtained which allows us to determine the invertible elements.