A5027830035- Image and Signal Denoising Methods
- Sparse and Compressive Sensing Techniques
- Medical Image Segmentation Techniques
- Advanced Numerical Analysis Techniques
- Mathematical Analysis and Transform Methods
- Photoacoustic and Ultrasonic Imaging
- Digital Filter Design and Implementation
- Medical Imaging Techniques and Applications
- Advanced Fluorescence Microscopy Techniques
- Cell Image Analysis Techniques
- Digital Holography and Microscopy
- Image Processing Techniques and Applications
- Advanced Image Fusion Techniques
- Advanced X-ray Imaging Techniques
- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Advanced MRI Techniques and Applications
- Numerical methods in inverse problems
- Advanced Electron Microscopy Techniques and Applications
- Optical measurement and interference techniques
- Seismic Imaging and Inversion Techniques
- Blind Source Separation Techniques
- Neural Networks and Applications
- Optical Coherence Tomography Applications
- Image Retrieval and Classification Techniques
École Polytechnique Fédérale de Lausanne
2016-2025
Centre d'Imagerie BioMedicale
2016-2025
Laboratoire d’Imagerie Biomédicale
2005-2023
Institute of Electrical and Electronics Engineers
2023
Signal Processing (United States)
2023
École Polytechnique
2004-2022
Ben-Gurion University of the Negev
2021
ORCID
2021
Bharathiar University
2020
SRM Institute of Science and Technology
2020
We present an automatic subpixel registration algorithm that minimizes the mean square intensity difference between a reference and test data set, which can be either images (two-dimensional) or volumes (three-dimensional). It uses explicit spline representation of in conjunction with processing, is based on coarse-to-fine iterative strategy (pyramid approach). The minimization performed according to new variation (ML*) Marquardt-Levenberg for nonlinear least-square optimization. geometric...
In this paper, we propose a novel deep convolutional neural network (CNN)-based algorithm for solving ill-posed inverse problems. Regularized iterative algorithms have emerged as the standard approach to problems in past few decades. These methods produce excellent results, but can be challenging deploy practice due factors including high computational cost of forward and adjoint operators difficulty hyper parameter selection. The starting point our work is observation that unrolled form CNN...
The article provides arguments in favor of an alternative approach that uses splines, which is equally justifiable on a theoretical basis, and offers many practical advantages. To reassure the reader who may be afraid to enter new territory, it emphasized one not losing anything because traditional theory retained as particular case (i.e., spline infinite degree). basic computational tools are also familiar signal processing audience (filters recursive algorithms), even though their use...
Abstract Background For the investigation of molecular mechanisms involved in neurite outgrowth and differentiation, accurate reproducible segmentation quantification neuronal processes are a prerequisite. To facilitate this task, we developed semiautomatic tracing technique. This article describes design validation Methods The technique was compared to fully manual delineation. Four observers repeatedly traced selected neurites 20 fluorescence microscopy images cells culture, using both...
This paper presents an account of the current state sampling, 50 years after Shannon's formulation sampling theorem. The emphasis is on regular where grid uniform. topic has benefitted from a strong research revival during past few years, thanks in part to mathematical connections that were made with wavelet theory. To introduce reader modern, Hilbert-space formulation, we reinterpret procedure as orthogonal projection onto subspace band-limited functions. We then extend standard paradigm...
This paper describes a new approach to the characterization of texture properties at multiple scales using wavelet transform. The analysis uses an overcomplete decomposition, which yields description that is translation invariant. It shown this representation constitutes tight frame l/sub 2/ and it has fast iterative algorithm. A characterized by set channel variances estimated output corresponding filter bank. Classification experiments with Brodatz textures indicate discrete (DWF) superior...
Mechanical properties of the adventitia are largely determined by organization collagen fibers. Measurements on waviness and orientation collagen, particularly at zero-stress state, necessary to relate structural mechanical response adventitia. Using fluorescence marker CNA38-OG488 confocal laser scanning microscopy, we imaged fibers in rabbit common carotid arteries ex vivo. The were cut open along their longitudinal axes get state. We used semi-manual automatic techniques measure...
The use of continuous B-spline representations for signal processing applications such as interpolation, differentiation, filtering, noise reduction, and data compressions is considered. coefficients are obtained through a linear transformation, which unlike other commonly used transforms space invariant can be implemented efficiently by filtering. same property also applies the indirect transform well evaluation approximating using smoothing or least squares splines. filters associated with...
Based on the theory of approximation, this paper presents a unified analysis interpolation and resampling techniques. An important issue is choice adequate basis functions. We show that, contrary to common belief, those that perform best are not interpolating. By opposition traditional interpolation, we call their use generalized interpolation; they involve prefiltering step when correctly applied. explain why approximation order inherent in any function limit artifacts. The decomposition...
We propose a new method for the intermodal registration of images using criterion known as mutual information. Our main contribution is an optimizer that we specifically designed this criterion. show well adapted to multiresolution approach because it typically converges in fewer evaluations than other optimizers. have built image pyramid, along with interpolation process, optimizer, and itself, around unifying concept spline-processing. This ensures coherence way model data yields good...
For pt.I see ibid., vol.41, no.2, p.821-33 (1993). A class of recursive filtering algorithms for the efficient implementation B-spline interpolation and approximation techniques is described. In terms simplicity realization reduction computational complexity, these compare favorably with conventional matrix approaches. interpretation (low-pass filter followed by an exact polynomial spline interpolator) smoothing least-squares methods proposed. These are applied to design digital filters...
This paper introduces a new approach to orthonormal wavelet image denoising. Instead of postulating statistical model for the coefficients, we directly parametrize denoising process as sum elementary nonlinear processes with unknown weights. We then minimize an estimate mean square error between clean and denoised one. The key point is that have at our disposal very accurate, statistically unbiased, MSE estimate-Stein's unbiased risk estimate-that depends on noisy alone, not Like MSE, this...
The sum and difference of two random variables with same variances are decorrelated define the principal axes their associated joint probability function. Therefore, histograms introduced as an alternative to usual co-occurrence matrices used for texture analysis. Two maximum likelihood classifiers presented depending on type object characterization (sum or some global measures). Experimental results indicate that conjointly nearly powerful cooccurrence discrimination. advantage proposed...
Efficient algorithms for the continuous representation of a discrete signal in terms B-splines (direct B-spline transform) and interpolative reconstruction (indirect with an expansion factor m are described. Expressions z-transforms sampled functions determined convolution property these kernels is established. It shown that both direct indirect spline transforms involve linear operators space invariant implemented efficiently by filtering. Fast computational based on recursive...
We present an algorithm for fast elastic multidimensional intensity-based image registration with a parametric model of the deformation. It is fully automatic in its default mode operation. In case hard real-world problems, it capable accepting expert hints form soft landmark constraints. Much fewer landmarks are needed and results far superior compared to pure registration. Particular attention has been paid factors influencing speed this algorithm. The B-spline deformation shown be...
We present a new, robust, computational procedure for tracking fluorescent markers in time-lapse microscopy. The algorithm is optimized finding the time-trajectory of single particles very noisy dynamic (two- or three-dimensional) image sequences. It proceeds three steps. First, images are aligned to compensate movement biological structure under investigation. Second, particle's signature enhanced by applying Mexican hat filter, which we show be optimal detector Gaussian-like spot 1/omega2...
We propose a general methodology (PURE-LET) to design and optimize wide class of transform-domain thresholding algorithms for denoising images corrupted by mixed Poisson-Gaussian noise. express the process as linear expansion thresholds (LET) that we relying on purely data-adaptive unbiased estimate mean-squared error (MSE), derived in non-Bayesian framework (PURE: risk estimate). provide practical approximation this theoretical MSE tractable optimization arbitrary thresholding. then...