Papers by Aleksandra Pizurica
Pattern Recognition Letters, 2006
Abstract We study the relationships between diffusivity functions in a nonlinear diffusion scheme... more Abstract We study the relationships between diffusivity functions in a nonlinear diffusion scheme and probabilities of edge presence under a marginal prior on ideal, noise-free image gradient. In particular we impose a Laplacian-shaped prior for the ideal gradient and we define the diffusivity function explicitly in terms of edge probabilities under this prior. The resulting diffusivity function has no free parameters to optimize. Our results demonstrate that the new diffusivity function, automatically, ie, without any parameter adjustments, satisfies ...
In packet switched networks, packets may get lost during transmission. As these networks are more... more In packet switched networks, packets may get lost during transmission. As these networks are more and more used for image and video communication, there is a growing need for efficient reconstruction algorithms. In wavelet coded images, the lost coefficients are typically replaced by zeros. This results in annoying black holes in the received image, mainly due to the loss of the low frequency content. In this chapter, we present a novel locally adaptive interpolation method for the reconstruction of the lost low frequency coefficients.
Abstract—In this paper we propose an improved video denoising scheme that combines recursive temp... more Abstract—In this paper we propose an improved video denoising scheme that combines recursive temporal filtering and wavelet domain spatial denoising. In our previous work, we introduced a sequential scheme (SEQWT) where a wavelet domain spatial filter is followed by a motion detector and by a selective recursive temporal filter. This scheme is efficient but its limitation is the lack of motion compensation.
In this paper we present an efficient way to both compute and extract salient information from tr... more In this paper we present an efficient way to both compute and extract salient information from trace transform signatures to perform object identification tasks. We also present a feature selection analysis of the classical trace-transform functionals, which reveals that most of them retrieve redundant information causing misleading similarity measurements. In order to overcome this problem, we propose a set of functionals based on Laguerre polynomials that return orthonormal signatures between these functionals.
abstract This paper presents a new method for unsupervised video segmentation based on mean shift... more abstract This paper presents a new method for unsupervised video segmentation based on mean shift clustering in spatio-temporal domain. The main novelties of the proposed approach are dynamic temporal adaptation of clusters due to which the segmentation evolves quickly and smoothly over time. The proposed method consists of a short initialization phase and an update phase. The proposed method significantly reduce the computation load for the mean shift clustering.
abstract Optical coherence tomography produces high resolution medical images based on spatial an... more abstract Optical coherence tomography produces high resolution medical images based on spatial and temporal coherence of the optical waves backscattered from the scanned tissue. However, the same coherence introduces speckle noise as well; this degrades the quality of acquired images. In this paper we propose a technique for noise reduction of 3D OCT images, where the 3D volume is considered as a sequence of 2D images, ie, 2D slices in depth-lateral projection plane.
abstract In this work we aim to improve the detection of Focal Cortical Dysplasia on MRI images u... more abstract In this work we aim to improve the detection of Focal Cortical Dysplasia on MRI images using a multimodal approach. We propose to estimate the thickness of the cortex jointly using partial volume maps of T1-weighted MPRAGE and T2-weighted FLAIR images by fitting spheres into the gray matter of the brain such that the amount of probability-weighted gray matter contained in each sphere is maximized.
Defined as the maximum amount of information which can be inserted in an original media for presc... more Defined as the maximum amount of information which can be inserted in an original media for prescribed transparency and robustness, watermarking capacity has been a challenging research topic in the last years. The present paper allows several current limitations in this respect to be overcame. As the capacity strongly depends on the attack statistical behaviour, the first part of our paper is devoted to their in-depth investigation.
Abstract—Recently, the NLMeans Filter has been proposed by Buades et al. for the suppression of w... more Abstract—Recently, the NLMeans Filter has been proposed by Buades et al. for the suppression of white Gaussian noise. This filter exploits the repetitive character of structures in an image, unlike conventional denoising algorithms, which typically operate in a local neighborhood. In this paper, we present an extension of this technique to the noise reduction of colored (correlated) noise, which is applicable in more realistic scenarios.
An overview is given of different state of the art modulator approaches useable as demodulating c... more An overview is given of different state of the art modulator approaches useable as demodulating component in Continuous Time-of-Flight range finding. The first is a spatial optical modulator which can be used in conjunction with a conventional image sensor; the second is a single bulk device combining both optical detection and modulation in the substrate.
Abstract—Preclinical in vivo micro computerized tomography suffers from high image noise, due to ... more Abstract—Preclinical in vivo micro computerized tomography suffers from high image noise, due to limitations on total scanning time and the small pixel sizes. A lot of different noise minimization algorithms have already been proposed to reconstruct images acquired in low dose settings. Sparse-view reconstruction amongst others can reduce acquisition dose significantly, by acquiring only a small subset of projection views. Total Variation minimization has been used extensively to solve these problems.
Computer vision (CV) is a diverse, relatively new and growing field in computer science. When usi... more Computer vision (CV) is a diverse, relatively new and growing field in computer science. When using CV for smart surveillance purposes, many questions arise: What do we want to watch after? What events are expected to happen? How to detect and label specific behaviours? Tracking the objects of interest is a crucial step before being able to answer such questions. After revisiting the state of the art of the techniques used in visual object tracking [1], we found many issues to be solved.
Digital imaging devices inevitably produce images corrupted with noise. The noise originates from... more Digital imaging devices inevitably produce images corrupted with noise. The noise originates from the sensors and analogue circuitry in the camera. In order to have better and sharper images and also for commercial reasons, there is a recent tendency to further increase the image resolution. Nowadays, cameras with more than 20 megapixels are not uncommon.
ABSTRACT In this paper, we propose a novel global Markov Random Field based image inpainting meth... more ABSTRACT In this paper, we propose a novel global Markov Random Field based image inpainting method with context-aware label selection. Context is determined based on the texture and color features in fixed image regions and is used to distinguish areas of similar content to which the search for candidate patches is limited.
abstract In this paper, we derive a novel robust image alignment technique that performs joint ge... more abstract In this paper, we derive a novel robust image alignment technique that performs joint geometric and photometric registration in the total least square sense. The main idea is to use the total least square metrics instead of the ordinary least square metrics, which is commonly used in the literature. While the OLS model indicates that the target image may contain noise and the reference image should be noise-free, this puts a severe limitation on practical registration problems.
ABSTRACT In this paper a denoising technique for digital gray value images corrupted with additiv... more ABSTRACT In this paper a denoising technique for digital gray value images corrupted with additive Gaussian noise is presented. We studied a recently proposed hard thresholding technique which uses a two stage selection procedure in which coefficients are selected based on their magnitude, spatial connectedness and interscale dependencies. We construct a shrinkage version of the algorithm which outperforms the original one.
Example-based super-resolution has become increasingly popular over the last few years for its ab... more Example-based super-resolution has become increasingly popular over the last few years for its ability to overcome the limitations of classical multi-frame approach. In this paper we present a new example-based method that uses the input low-resolution image itself as a search space for high-resolution patches by exploiting self-similarity across different resolution scales. Found examples are combined in a high-resolution image by the means of Markov Random Field modelling that forces their global agreement.
Abstract—Vascular image registration is an important phase in the diagnosis and treatment of medi... more Abstract—Vascular image registration is an important phase in the diagnosis and treatment of medical conditions. Images obtained from different scanners and at different times need to be aligned in order to diagnose strokes and to provide a full insight into a head model for planning the operation. An inexact graph matching solution is developed on the brain blood vessel skeletons obtained after segmentation and skeletonization.
abstract In this paper we present a new method for superresolution of depth video sequences using... more abstract In this paper we present a new method for superresolution of depth video sequences using high resolution color video. Here we assume that the depth sequence does not contain outlier points which can be present in the depth images. Our method is based on multiresolution decomposition, and uses multiple frames to search for a most similar depth segments to improve the resolution of the current frame. First step is the wavelet decomposition of both color and depth images.
Uploads
Papers by Aleksandra Pizurica