A novel super-resolution reconstruction (SRR) framework in magnetic resonance imaging (MRI) is proposed. Its purpose is to produce images of both high resolution and high contrast desirable for image-guided minimally invasive brain surgery. The input data are multiple 2-D multislice inversion recovery MRI scans acquired at orientations with regular angular spacing rotated around a common frequency encoding axis. The output is a 3-D volume of isotropic high resolution. The inversion process resembles a localized projection reconstruction problem. Iterative algorithms for reconstruction are based on the projection onto convex sets (POCS) formalism. Results demonstrate resolution enhancement in simulated phantom studies, and ex vivo and in vivo human brain scans, carried out on clinical scanners. A comparison with previously published SRR methods shows favorable characteristics in the proposed approach.

Image segmentation or image classification is very often turned into a problem of energy minimisation. Thus, the energy functional is defined as the summation of a data term, which defines the properties of the object to be segmented with regard to the image, and a prior term, which integrates the knowledge that we have about the object. The definition and representation of the prior term appear to be critical for several reasons, among which: should the prior constraint be generic or specific, and how to generate it? what tradeoff between sophisticated knowledge and computational efficiency? We address in this paper the core problem of local minima which may trap the active contour, thus preventing it from finding the object of interest in the image. We propose to incorporate a prior shape constraint which weight is a function of space and time. We will show experimentally that a relaxed constraint at the begining of …

In this paper we propose a novel approach for updating the building layer of a high-scale land use digital map using a more recent high resolution panchromatic Quickbird image. A preliminary map-to-image fine matching is carried out by edge-based and shape constrained active contours. This step reduces exogenous discrepancies between the map and the image and makes a subsequent Hough voting change detection reliable. Promising experimental results are shown over Beijing city area.

This paper exposes a novel formulation of prior shape constraint incorporation for the level set segmentation of objects from corrupted images. Applicable to variational frameworks, the proposed scheme consists in weighting the prior shape constraint by a function of time and space to overcome local minima issues of the energy functional. Pose parameters which make the prior shape constraint invariant from global transformations are estimated by the downhill simplex algorithm, which is more tractable and robust than the traditional gradient descent. The proposed scheme is simple, easy to implement and can be generalized to any variational approach incorporating a single prior shape. Results illustrated with different kinds of images demonstrate the efficiency of the method.

We propose a novel approach for digital building map refinement based on the use of knowledge-driven active contours and very high resolution panchromatic optical imagery. This methodology is designed to finely match each building symbolized in an urban Geographical Information System (GIS) database onto its counterpart representation in remote sensing data. This method is GIS mapdriven: GIS data globally registered to the image allows to initialize an active contour near the target building in the image to achieve subsequent refinement. Moreover the digital map provides valuable shape information about the object in the image we aim at matching. This geometric and specific prior knowledge is embedded as a shape constraint into the active contour and enables to overcome urban artifacts issues. Besides, we propose to embed a coarse Digital Surface Model (DSM) as well as a spatio-temporal shape prior constraint within the active contour model. Experimental results carried out over Beijing city area and illustrated in this paper show how these latter contributions improve the robustness and speed of the map refinement process. Map refinement addressed in this paper is becoming an essential issue for urban planning, telecommunications, automobile navigation, crisis and pollution management, which all rely on up-to-date and precise digital maps of a city.

In this paper, we propose two methods aiming at updating geographic information system (GIS) urban maps using satellite high-resolution remote sensed images. Both methods are based on the input of a priori knowledge provided by GIS data, and a digital surface model (DSM). This article introduces theoretical aspects of our methods.

In this paper we propose a novel approach for updating the building layer of a high-scale land use digital map using a more recent high resolution panchromatic Quickbird image. A preliminary map-to-image fine matching is carried out by edge-based and shape constrained active contours. This step reduces exogenous discrepancies between the map and the image and makes a subsequent Hough voting change detection reliable. Promising experimental results are shown over Beijing city area.