Kernel-based methods for change detection in remote sensing images

Nowadays, the joint exploitation of images acquired daily by remote sensing instruments and of images available from archives allows a detailed monitoring of the transitions occurring at the surface of the Earth. These modifications of the land cover generate spectral discrepancies that can be detected via the analysis of remote sensing images. Independently from the origin of the images and of type of surface change, a correct processing of such data implies the adoption of flexible, robust and possibly nonlinear method, to correctly account for the complex statistical relationships characterizing the pixels of the images. This Thesis deals with the development and the application of advanced statistical methods for multi-temporal optical remote sensing image processing tasks. Three different families of machine learning models have been explored and fundamental solutions for change detection problems are provided. In the first part, change detection with user supervision has been considered. In a first application, a nonlinear classifier has been applied with the intent of precisely delineating flooded regions from a pair of images. In a second case study, the spatial context of each pixel has been injected into another nonlinear classifier to obtain a precise mapping of new urban structures. In both cases, the user provides the classifier with examples of what he believes has changed or not. In the second part, a completely automatic and unsupervised method for precise binary detection of changes has been proposed. The technique allows a very accurate mapping without any user intervention, resulting particularly useful when readiness and reaction times of the system are a crucial constraint. In the third, the problem of statistical distributions shifting between acquisitions is studied. Two approaches to transform the couple of bi-temporal images and reduce their differences unrelated to changes in land cover are studied. The methods align the distributions of the images, so that the pixel-wise comparison could be carried out with higher accuracy. Furthermore, the second method can deal with images from different sensors, no matter the dimensionality of the data nor the spectral information content. This opens the doors to possible solutions for a crucial problem in the field: detecting changes when the images have been acquired by two different sensors.

The Smartcanula: a new tool for remote access perfusion in limited access cardiac surgery.

Devices for venous cannulation have seen significant progress over time: the original, rigid steel cannulas have evolved toward flexible plastic cannulas with wire support that prevents kinking, very thin walled wire wound cannulas allowing for percutaneous application, and all sorts of combinations. In contrast to all these rectilinear venous cannula designs, which present the same cross-sectional area over their entire intravascular path, the smartcanula concept of "collapsed insertion and expansion in situ" is the logical next step for venous access. Automatically adjusting cross-sectional area up to a pre-determined diameter or the vessel lumen provides optimal flow and ease of use for both, insertion and removal. Smartcanula performance was assessed in a small series of patients (76 +/- 17 kg) undergoing redo procedures. The calculated target pump flow (2.4 L/min/m2) was 4.42 +/- 61 L/ min. Mean pump flow achieved during cardiopulmonary bypass was 4.84 +/- 87 L/min or 110% of the target. Reduced atrial chatter, kink resistance in situ, and improved blood drainage despite smaller access orifice size, are the most striking advantages of this new device. The benefits of smart cannulation are obvious in remote cannulation for limited access cardiac surgery, but there are many other cannula applications where space is an issue, and that is where smart cannulation is most effective.

Structural characterization of rockfall sources in Yosemite Valley from remote sensing data and field surveys

Yosemite Valley poses significant rockfall hazard and related risk due to its glacially steepened walls and approximately 4 million visitors annually. To assess rockfall hazard, it is necessary to evaluate the geologic structure that contributes to the destabilization of rockfall sources and locate the most probable future source areas. Coupling new remote sensing techniques (Terrestrial Laser Scanning, Aerial Laser Scanning) and traditional field surveys, we investigated the regional geologic and structural setting, the orientation of the primary discontinuity sets for large areas of Yosemite Valley, and the specific discontinuity sets present at active rockfall sources. This information, combined with better understanding of the geologic processes that contribute to the progressive destabilization and triggering of granitic rock slabs, contributes to a more accurate rockfall susceptibility assessment for Yosemite Valley and elsewhere.

Remote control of pulmonary blood flow: a dream comes true.

The indication for pulmonary artery banding is currently limited by several factors. Previous attempts have failed to produce adjustable pulmonary artery banding with reliable external regulation. An implantable, telemetrically controlled, battery-free device (FloWatch) developed by EndoArt SA, a medical company established in Lausanne, Switzerland, for externally adjustable pulmonary artery banding was evaluated on minipigs and proved to be effective for up to 6 months. The first human implant was performed on a girl with complete atrioventricular septal defect with unbalanced ventricles, large patent ductus arteriosus and pulmonary hypertension. At one month of age she underwent closure of the patent ductus arteriosus and FloWatch implantation around the pulmonary artery through conventional left thoracotomy. The surgical procedure was rapid and uneventful. During the entire postoperative period bedside adjustments (narrowing or release of pulmonary artery banding with echocardiographic assessment) were repeatedly required to maintain an adequate pressure gradient. The early clinical results demonstrated the clinical benefits of unlimited external telemetric adjustments. The next step will be a multi-centre clinical trial to confirm the early results and adapt therapeutic strategies to this promising technology.

Variation in performance across Swiss mammography screening programmes: the influence of reading strategy and experience.

[Abstract] Reading volume and mammography screening performance appear positively correlated. Performance was compared across organised Swiss screening programmes, which target relatively small populations. Except for accreditation of 2nd readers radiologists (restrictive vs non-restrictive strategy), Swiss programmes have similar screening regimen/procedures and duration, which maximises comparability. Variation in performance was explored in order to improve mammography practice and optimise screening performance. Indicators of quality and effectiveness were evaluated for about 200,000 screens performed over 4 screening rounds in the 3 longest-standing Swiss cantonal programmes (of Vaud, Geneva and Valais). Interval cancers were identified by linkage with cancer registries records. Most European standards of performance were met with a favourable cancer stage shift. Several performance indicators showed substantial variation across programmes. In subsequent rounds, compared with programmes (Vaud and Geneva) which accredited few 2nd readers to increase their individual reading volume, proportions of in situ lesions and of small cancers (? 1cm) were one third lower and halved, respectively, and the proportion of advanced lesions (stage II+) nearly 50% higher in the programme without a restrictive selection strategy. Discrepancy in second-year proportional incidence of interval cancers appears to be multicausal. Differences in performance could partly be explained by a selective strategy for 2nd readers and a prior experience in service screening, but not by the levels of opportunistic screening and programme attendance. This study provides clues for enhancing mammography screening performance in low-volume Swiss programmes.

Active learning methods for remote sensing image classification

In this paper, we propose two active learning algorithms for semiautomatic definition of training samples in remote sensing image classification. Based on predefined heuristics, the classifier ranks the unlabeled pixels and automatically chooses those that are considered the most valuable for its improvement. Once the pixels have been selected, the analyst labels them manually and the process is iterated. Starting with a small and nonoptimal training set, the model itself builds the optimal set of samples which minimizes the classification error. We have applied the proposed algorithms to a variety of remote sensing data, including very high resolution and hyperspectral images, using support vector machines. Experimental results confirm the consistency of the methods. The required number of training samples can be reduced to 10% using the methods proposed, reaching the same level of accuracy as larger data sets. A comparison with a state-of-the-art active learning method, margin sampling, is provided, highlighting advantages of the methods proposed. The effect of spatial resolution and separability of the classes on the quality of the selection of pixels is also discussed.

Semi-supervised remote sensing image classification with cluster kernels

A semisupervised support vector machine is presented for the classification of remote sensing images. The method exploits the wealth of unlabeled samples for regularizing the training kernel representation locally by means of cluster kernels. The method learns a suitable kernel directly from the image and thus avoids assuming a priori signal relations by using a predefined kernel structure. Good results are obtained in image classification examples when few labeled samples are available. The method scales almost linearly with the number of unlabeled samples and provides out-of-sample predictions.

Domain Adaptation in remote sensing: increasing the portability of land-cover classifiers

Among the types of remote sensing acquisitions, optical images are certainly one of the most widely relied upon data sources for Earth observation. They provide detailed measurements of the electromagnetic radiation reflected or emitted by each pixel in the scene. Through a process termed supervised land-cover classification, this allows to automatically yet accurately distinguish objects at the surface of our planet. In this respect, when producing a land-cover map of the surveyed area, the availability of training examples representative of each thematic class is crucial for the success of the classification procedure. However, in real applications, due to several constraints on the sample collection process, labeled pixels are usually scarce. When analyzing an image for which those key samples are unavailable, a viable solution consists in resorting to the ground truth data of other previously acquired images. This option is attractive but several factors such as atmospheric, ground and acquisition conditions can cause radiometric differences between the images, hindering therefore the transfer of knowledge from one image to another. The goal of this Thesis is to supply remote sensing image analysts with suitable processing techniques to ensure a robust portability of the classification models across different images. The ultimate purpose is to map the land-cover classes over large spatial and temporal extents with minimal ground information. To overcome, or simply quantify, the observed shifts in the statistical distribution of the spectra of the materials, we study four approaches issued from the field of machine learning. First, we propose a strategy to intelligently sample the image of interest to collect the labels only in correspondence of the most useful pixels. This iterative routine is based on a constant evaluation of the pertinence to the new image of the initial training data actually belonging to a different image. Second, an approach to reduce the radiometric differences among the images by projecting the respective pixels in a common new data space is presented. We analyze a kernel-based feature extraction framework suited for such problems, showing that, after this relative normalization, the cross-image generalization abilities of a classifier are highly increased. Third, we test a new data-driven measure of distance between probability distributions to assess the distortions caused by differences in the acquisition geometry affecting series of multi-angle images. Also, we gauge the portability of classification models through the sequences. In both exercises, the efficacy of classic physically- and statistically-based normalization methods is discussed. Finally, we explore a new family of approaches based on sparse representations of the samples to reciprocally convert the data space of two images. The projection function bridging the images allows a synthesis of new pixels with more similar characteristics ultimately facilitating the land-cover mapping across images.