Image Mosaicing

and key words This paper deals with the calibration of a cylindrical omnidirectional imaging system, based on a rotating 2048 pixels linear camera which provides high definition panoramas. The two-step algorithm relies on line segment projections as calibration features, which are sinusoidal curves. A cylindrical image line detection algorithm is proposed, based on the dual principle of the Hough transform. Moreover, the use of Plucker coordinates introduces some new characteristics in the calibration process. This kind of formalism allows a linearization of the cylindrical projection, which is non-linear in the usual way. Results obtained from this first step are used to evaluate one of the intrinsic parameters, the other one being determined by a linear criterion minimization in the dual space, ie the sine magnitude space. Non-Linear Calibration, Omnidirectional Imaging System, Plucker, Coordinates, Hough Transform.

Nous présentons ici la géométrie d'un capteur panoramique tournant, ainsi que la généralisation du concept de contrainte épipolaire sur des images de type cylindrique. Différents procédés de minimisation sont utilisés pour retrouver certains paramètres expérimentaux. Cette contrainte permet ainsi d'optimiser de l'étape de recherche de points appariés afin de reconstituer la configuration tridimensionnelle de la scène observée à partir d'images cylindriques haute définition.

Fetoscopic laser photo-coagulation of the placental vascular anastomoses remains the most effective therapy for twin-to-twin transfusion syndrome (TTTS) in monochorionic twin pregnancies. However, to ensure the success of the intervention, complete photo-coagulation of all anastomoses is needed. This is made difficult by the limited field of view of the fetoscopic video guidance, which hinders the surgeon's ability to locate all the anastomoses. A potential solution to this problem is to expand the field of view of the placental surface by creating a mosaic from overlapping fetoscopic images. This mosaic can then be used for anastomoses localization and spatial orientation during surgery. However, this requires accurate and fast algorithms that can operate within the real-time constraints of fetal surgery. In this work, we present an image mosaicing framework that leverages the parallelism of modern GPUs and can process clinical fetoscopic images in real-time. Initial qualitative results on ex-vivo placental images indicate that the proposed framework can generate clinically useful mosaics from fetoscopic videos in real-time.

The Mercury Dual Imaging System (MDIS) on the MESSENGER spacecraft will provide critical measurements tracing Mercury's origin and evolution. MDIS consists of a monochrome narrow-angle camera (NAC) and a multispectral wide-angle camera (WAC). The NAC is a 1.5°field-of-view (FOV) off-axis reflector, coaligned with the WAC, a fourelement refractor with a 10.5°FOV and 12-color filter wheel. The focal plane electronics of each camera are identical and use a 1,024 × 1,024 Atmel (Thomson) TH7888A chargecoupled device detector. Only one camera operates at a time, allowing them to share a common set of control electronics. The NAC and the WAC are mounted on a pivoting platform that provides a 90°field-of-regard, extending 40°sunward and 50°anti-sunward from the spacecraft +Z-axis-the boresight direction of most of MESSENGER's instruments. Onboard data compression provides capabilities for pixel binning, remapping of 12-bit data into 8 bits, and lossless or lossy compression. MDIS will acquire four main data sets at Mercury during three flybys and the two-Mercury-solar-day nominal mission: a monochrome

A program is underway to digitize all mercurian craters greater than 5 km in diameter on Mariner 10 photography1. The method is very 'similar to that used to collect datafor the Brown University Mars Crater ~i b r a r y ~. Additionally, an independent catalog of all nearside lunar craters greater than 7 km in diameter has been completed recently at the University of Arizona Lunar and Planetary ~aboratory3. These catalogs include data on crater location, diameter, morphology, and degradation. In addition, both the mercurian and lunar catalogs contain morphometric data on central peak heights and crater depths. These catalogs, yielding data for over 65,000 craters on the three planets, provide comprehensive bases for detailed crater population studies4. Comparison with previous work -Past studies of lunar5, mercurian6, a d mar-

In this paper we show how based on a number of different techniques it is possible to fully automatically generate basic ingredients for high quality visualizations of urban areas characterized by vertical facade planes from nothing but uncalibrated wide-baseline image sequences without using any markers or ground control. At the core of our algorithms are least-squares matching, projective geometry based reconstruction, robust estimation based on random sample consensus -RANSAC, auto-calibration, projective and Euclidean bundle adjustment, plane to plane homographies, as well as the robust estimation of image mosaics. Results for the Hradschin in Prague, Czechia, and Plaza Real in Barcelona, Spain, show the potential and shortcomings of the employed algorithms.

In this paper we show how based on a number of different techniques it is possible to fully automatically generate basic ingredients for high quality visualizations of urban areas characterized by vertical facade planes from nothing but uncalibrated wide-baseline image sequences without using any markers or ground control. At the core of our algorithms are least-squares matching, projective geometry based reconstruction, robust estimation based on random sample consensus -RANSAC, auto-calibration, projective and Euclidean bundle adjustment, plane to plane homographies, as well as the robust estimation of image mosaics. Results for the Hradschin in Prague, Czechia, and Plaza Real in Barcelona, Spain, show the potential and shortcomings of the employed algorithms.

Laboratory made granular-granular impact craters have been used as model analogues of planetary impact craters. These kind of craters have been observed and studied using profilometry techniques that allow to retrieve important morphologic features from the impacted surface. In this work, we propose to use a Time-of-Flight camera (Microsoft Kinect One) for the acquisition of depth data. We show comparisons between the typically used technique and the analysis derived from the Time-of-Flight data. We also release craterslab, a Python library developed to automate most of the tasks from the process of studying impact craters produced by granular projectiles hitting on the surface of granular targets. The library is able to acquire, identify, and measure morphological features of impacted surfaces through the reconstruction of 3D topographic maps. Our results show that using a Time-of-Flight camera and automating the data processing with a software library for the systematic study of impact craters can produce very accurate results while reducing the time spent on different stages of the process.

An important part of our knowledge is in the form of images. For example, a large amount of geophysical and environmental data comes from satellite photos, a large amount of the information stored on the Web is in the form of images, etc. It is therefore desirable to use this image information in data mining. Unfortunately, most existing data mining techniques have been designed for mining numerical data and are thus not well suited for image databases. Hence, new methods are needed for image mining. In this paper, we show how data mining can be used to find common patterns in several images.

This paper presents a new error equalisation method for construction of mosaics built from a large number of images that may not completely cover a scene ('sparse' coverage without all the overlapping images in the neighbourhood). The proposed method is shown to achieve consistent sparse image mosaic construction with the final alignment error no bigger, and in most cases smaller, than the error introduced by the transformation parameter estimation method. The performance of the proposed method is validated using simulation data as well as X-ray images acquired from non-destructive inspection of physically large aircraft components.

Este estudo teve como objetivo identificar e caracterizar as formas e os processos fluviais dominantes associados ao padrão de canal meandrante no vale aluvial do baixo curso do rio do Peixe, estado de São Paulo. As interpretações geomorfológicas foram baseadas em imagens orbitais, fotografias de sobrevoo, dados hidrológicos e levantamentos em campo. No vale aluvial do baixo curso do rio do Peixe foram caracterizadas unidades de terraços, planície de inundação e canais, ordenados como 1º táxon. Nessas unidades encontram-se sub-unidades denominadas de unidades geomórficas, como por exemplo, na planície de inundação ocorrem paleocanais, lagos em ferradura e bacias de inundação, estas últimas, portanto ordenadas como 2 º táxon. Também foram descritos como de 2 º táxon a faixa de meandros, leque de espraiamento e leque aluvial. Na unidade do canal destacam-se as barras fluviais, principalmente, as barras em pontal, características dos rios meandrantes, além de barras centrais e laterais...

We introduce and compare two algorithms related to ego-motion, applicable to a robot using a panoramic visual sensor in an unknown environment. The first method, computationally cheap, extends a family of bio-inspired navigation systems by providing an orientation estimation that previously had to be obtained from external references. The second method solves the general structurefrom-motion problem and is here adapted to the panoramic sensor case. Both methods produce good results in a test scene, though the calibrated method has a greater precision. A step-by-step comparison of the methods is included, with mobile robotics applications in mind.

Canada's RADARSAT -1 satellite has accomplished several systematically planned coverage campaigns for Africa in the past nine years of satellite operations under the Canadian Space Agency's 'Background Mission'. The Mission is about creating unifor m Synthetic Aperture Radar (SAR) data archives to serve as global benchmarks for change monitoring, land mapping and disaster management. The first coverage of Africa was completed in 1997 for a rapid overview of the continent and was conducted using the wide area ScanSAR imaging beam of the satellite. Another continent-wide coverage was undertaken a few years later by using 25 m resolution Wide beam for the sub-Saharan Africa and 50 m resolution ScanSAR Narrow beam for the Saharan and the northern portion of the continent. In parallel, and as part of the baseline data acquisition objectives of the satellite mission, 25 m resolution radargrammetric data were collected over the whole of Africa by making use of the uniqu...

As evidence for the wide and continuing interest in astronomical software issues, 304 participants attended the conference from 21 countries. The theme of "enabling technologies" was addressed by A. Szalay (reviewing the genesis, challenges, and opportunities of the virtual observatory), J. Spyromilio (commissioning the VLT), and J. Tarter and D. Werthimer (the science and technical challenges of the SETI project). S. Murray and R. Brissenden described the early scientific successes and operational systems for the Chandra X-Ray Observatory, and B. Glendenning gave an overview of the software engineering challenges facing the ALMA telescope project. Key papers on software development technologies included J. Graybeal (CORBA), G. Filippi (software development methodologies employed at ESO), C. Kesselman (computational grids), J. Manuel Filgueira (distributed objects), and V. Yodaiken (real-time Linux). In the area of science data processing pipelines, R. Lupton described the image reduction pipeline for the Sloan Digital Sky Survey, and R. Cutri described the 2MASS project under the topic of sky surveys. K. Shortridge gave an insightful and amusing historical perspective of astronomical software development. The oral program also included a number of excellent contributed papers, and daily poster sessions (with over 120 posters presented in total) included presentations on all areas of software development and systems. Fourteen groups demonstrated their latest software systems and astronomical information services. Over the past decade the emphasis of the conference has gradually changed, from the "Data Analysis" to the "Systems" part of the title. This change reflects the fact that astronomical software development has entered into the age of very large projects, either in support of large, complex, forefront telescopes (Gemini, VLT, ALMA, and NGST for example) or in the support of massive surveys and the associated terabytescale databases (e.g., 2MASS, GSC-II, and SDSS). The soft-

In this paper, a novel algorithm for creating virtual indoor environments is described. First, a panoramic mosaic is generated from a series of photos taken with a camera rotates along a horizontal axis. Then, from the panoramic mosaic image, a non-...

Mid-infrared observations of the Andromeda galaxy, M31, obtained with the Infrared Array Camera on board the Spitzer Space Telescope are presented. The image mosaics cover areas of approximately 3. • 7 × 1. • 6 and include the satellite galaxies M32 and NGC 205. The appearance of M31 varies dramatically in the different mid-infrared bands, from the smooth bulge and disk of the old stellar population seen at 3.6 µm to the well-known '10 kpc ring' dominating the 8 µm image. The similarity of the 3.6 µm and optical isophotes and nearly constant optical-mid-infrared color over the inner 400 ′′ confirms that there is no significant extinction at optical wavelengths in M31's bulge. The nuclear colors indicate the presence of dust but not an infrared-bright active galactic nucleus. The integrated 8 µm non-stellar luminosity implies a star formation rate of 0.4M ⊙ yr -1 , consistent with other indicators that show M31 to be a quiescent galaxy.

L'objectif est de proposer une méthodologie complète de reconstruction 3D d'objets sous-marins à des fins de mesures quantitatives. Afin de faciliter le calcul de la reconstruction 3D, nous utilisons un asservissement visuel à l'aide d'un système de stéréovision permettant d'obtenir différentes vues de l'objet à intervalles réguliers selon une trajectoire prédéfinie.

This document presents the objectives, the work done and the results for the Habitat Mapping Scenario of the Azores FREEsubNET workshop, held in Horta from 19 to 23 July 2010. Three algorithms were tested, two of which have been developed in the FREEsubNET network, for the purpose of image mosaicking of two chosen sites: the Espalamaca hydrothermal vents and a small zone inside the Monte da Guia marine reserve. The former was very recently discovered and is very important from the biological point of view. The test was challenging but the results validated the developed algorithms. In addition, data from a mission with a multibeam sonar in the Espalamaca site are presented in order to be fused with the optical maps.

Le travail présenté dans cet article concerne l'optimisation du contrôle géométrique des tubes cintrés, utilisés dans le secteur automobile. Cette optimisation vise à minimiser les écarts géométriques moyens, déterminés par ce contrôle et influencés par les positions de trois bras magnétiques. Tout d'abord, nous avons utilisé un système expérimental et un plan factoriel complet de trois facteurs à deux niveaux pour effectuer les expériences. Ensuite, les réponses en fonction des facteurs et leurs interactions ont été modélisées par une régression multiple. Les modèles développés ont des coefficients de détermination supérieurs à 90%. Finalement, les positions optimales des trois bras magnétiques sont déterminées par un diagramme d'optimisation.

RESOLUTION IMAGING V. J. Bray, O. L. White, K. N. Singer, W. B. McKinnon, P. M. Schenk, S. J. Robbins, J. M. Moore, R. D. Dhingra, J. R. Spencer, C. B. Olkin, J. W. Parker, S. A. Stern, A. J. Verbiscer, H. A. Weaver, and The New Horizons Geology, Geophysics and Imaging (GGI) Team. 1. Lunar and Planetary Lab., The University of Arizona, Tucson, AZ, 85721. SETI Institute, Mountain View, CA. Southwest Research Inst., Boulder, CO. Washington U. in St. Louis, MO. Lunar and Planetary Inst., Houston, TX. NASA AmesResearch Center, Moffett Field, CA. University of Idaho, Perimeter Drive, Moscow, ID. University of Virginia, Charlottesville, VA. JHU Applied Physics Lab., Laurel, MD. (vjbray@lpl.arizona.edu).

Hyperspectral imagery contains many different frequency bands for each pixel, enabling advanced detection algorithms. The so called pushbroom sensor technology is the most widespread realization of hyperspectral camera. Such cameras capture a single row of pixels, but each pixel can have hundreds of associated frequency bands. Assembling a full image from these pixel rows typically requires having a precise knowledge of the movement of the camera in space while it scans the desired area. In this research we present a procedure to build hyperspectral images from line sensor data without camera position information. The approach relies on an accompanying conventional camera and exploits the homographies between images of planar scenes for image mosaic construction. The hyperspectral camera is geometrically calibrated by using a special target and a variation of the Direct Linear Transform algorithm. This enables mapping the hyperspectral lines to each of the images in the mosaic, and a hyperspectral image can be built using the same homographies computed for the mosaic. Ideally, this research can enable small UAVs to perform enhanced detection and navigation previously reserved for high altitude aircraft.

L'etude porte sur un Systeme Stereo Catadioptrique multiPlan (SSCP) constitue d'une unique camera perspective et deux miroirs plans judicieusement positionnes. En positionnant un objet entre les miroirs plans et la camera, cinq vues en sont observees simultanement. Dans cet article, nous abordons la modelisation et l'etalonnage geometrique de SSCP. Les points d'interet detectes dans la vue directe servent de reference pour la recherche de correspondances sous la contrainte d'une homographie minimisant l'erreur photometrique entre voisinages. La correspondance mene a la reconstruction 3D des points dont les resultats demontrent la qualite de la methode de recherche de correspondances et que la reconstruction 3D avec une image d'un SSCP est possible.

We have used the 28 ′ × 28 ′ Hubble Space Telescope image mosaic from the GEMS (Galaxy Evolution from Morphology and SEDs) survey in conjunction with the COMBO-17 photometric redshift survey to constrain the incidence of major mergers between spheroid-dominated galaxies with little cold gas (dry mergers) since z = 0.7. A set of N-body merger simulations was used to explore the morphological signatures of such interactions: they are recognizable either as < 5 kpc separation close pairs or because of broad, low surface brightness tidal features and asymmetries. Data with the depth and resolution of GEMS are sensitive to dry mergers between galaxies with M V −20.5 for z 0.7; dry mergers at higher redshifts are not easily recovered in single-orbit HST imaging. Six dry mergers (12 galaxies) with luminosity ratios between 1:1 and 4:1 were found from a sample of 379 red early-type galaxies with M V < −20.5 and 0.1 < z < 0.7. The simulations suggest that the morphological signatures of dry merging are visible for ∼ 150 Myr and we use this timescale to convert the observed merger incidence into a rate. On this basis we find that present day spheroidal galaxies with M V < −20.5 on average have undergone between 0.5 and 2 major dry mergers since z ∼ 0.7. We have compared this result with the predictions of a Cold Dark Matter based semi-analytic galaxy formation model. The model reproduces the observed declining major merger fraction of bright galaxies and the space density of luminous early-type galaxies reasonably well. The predicted dry merger fraction is consistent with our observational result. Hence, hierarchical models predict and observations now show that major dry mergers are an important driver of the evolution of massive early-type galaxies in recent epochs.

We have used the 30 ′ × 30 ′ Hubble Space Telescope image mosaic from the Galaxy Evolution from Morphology and SEDs (GEMS) project in conjunction with the COMBO-17 deep photometric redshift survey to define a sample of nearly 1500 galaxies with 0.65 ≤ z ≤ 0.75. With this sample, we can study the distribution of rest-frame V-band morphologies more than 6 Gyr ago, without differential bandpass shifting and surface brightness dimming across this narrow redshift slice. Focusing on red-sequence galaxies at z ∼ 0.7, we find that 85% of their combined restframe V-band luminosity density comes from visually-classified E/S0/Sa galaxies down to M V − 5 log 10 h −19.5. Similar results are obtained if automated classifiers are used. This fraction is identical to that found at the present day, and is biased by less than 10% by large scale structure and the morphology-density relation. Under the assumption that peculiar and edge-on disk galaxies are red by virtue of their dust content, we find that less than 13% of the total rest-frame V-band luminosity of the z ∼ 0.7 red galaxy population is from dusty galaxies.

We propose a new incremental motion estimation algorithm to deal with long image sequences. It applies to a sliding window of triplets of images, but unlike previous approaches, which rely on point matches across three or more views, we also use those points shared only by two views. This is important because matches between two views are more common than those across more views. The problem is formulated as a series of local bundle adjustments in such a way that the estimated camera motions in the whole sequence are consistent with each other. Two implementations are described. The first is an exact one, which, based on the observation of the sparse structure of the adjustment network, embeds the optimization of 3D structure parameters within the optimization of the camera pose parameters. This optimization embedding considerably reduces the minimization complexity. The second is a mathematical procedure which transforms the original problem involving both 3D structure and pose parameters into a much smaller one, a minimization over just the camera's pose parameters. This leads to an even higher computational gain. Because we make full use of local image information, our technique is more accurate than previous incremental techniques, and is very close to, and considerably faster than, global bundle adjustment. Experiments with both synthetic and real data have been conducted to compare the proposed technique with other techniques, and have shown our technique to be clearly superior.

Classic mosaic is one of the oldest and most durable art forms. There has been a growing interest in simulating classic mosaics from digital images recently. To be visually pleasing, a mosaic should satisfy the following constraints: tiles should be non-overlapping, tiles should align to the perceptually important edges in the underlying digital image, and orientation of the neighbouring tiles should vary smoothly across the mosaic. Most of the existing approaches operate in two steps: first they generate tile orientation field and then pack the tiles according to this field. However, previous methods perform these two steps based on heuristics or local optimisation which, in some cases, is not guaranteed to converge. Some other major disadvantages of previous approaches are: (i) either substantial user interaction or hard decision making such as edge detection is required before mosaicing starts (ii) the number of tiles per mosaic must be fixed beforehand, which may cause either undesired overlap or gap space between the tiles. In this work, we propose a novel approach by formulating the mosaic simulating problem in a global energy optimisation framework. Our algorithm also follows the two-step approach, but each step is performed with global optimisation. For the first step, we observe that the tile orientation constraints can be naturally formulated in an energy function that can be optimised with the α-expansion algorithm. For the second step of tightly packing the tiles, we develop a novel graph cuts based algorithm. Our approach does not require user interaction, explicit edge detection, or fixing the number of tiles, while producing results that are visually pleasing.

This paper describes a system for high resolution video conferencing. A number of camcorders are used to capture the video, which are then mosaiced to generate a wide angle panoramic view. Furthermore this system is made "real-time" by detecting changes and updating them on the mosaic. This system can be deployed on a single machine or on a cluster for better performance. It is also scalable and shows a good real-time performance. The main application for this system is videoconferencing for distance learning but it can be used for any high resolution broadcasting.

This paper introduces a novel benchmarking tool for measuring the robustness of existing mosaicing algorithms in presence of a given set of disturbances. The process combines a set of partially overlapping images into a wide-view result used to represent UAV image series and orthophotography. Geometrical misalignements caused by perspective error may lead to some unpredictable artifacts and phantom effects in the mosaics. A very few solutions measure their immunity to known distortions and mainly focus on registration accuracy measurement. Only limited attention was given to characterize the response of the actual image fusion algorithms and their capacity to properly preserve content geometries. In this paper, we also introduce a new fidelity metric assessing the mosaicing response to a disturbance of a given extent used as prior information. The metric helps to better define the use cases fulfilling aerial imaging requirements.

There are situations where it is not possible to capture or scan a large document with given imaging media such as Xerox machine or scanner as a single image in a single exposure because of their inherent limitations. This results in capturing or scanning of large document into number of split components of a document. Hence, there is a lot of scope for mosaicing the several split images into a single large document image. In this work, we present a novel technique Fourier Transform (FT) based Column-Block (CB) and Row-Block (RB) matching procedure to mosaic the two split images of a large document in order to build an original and single large document image. The FT is rarely used in the analysis of documents since it provides only the global information of the document. The global information doesn't help in analyzing the documents of split images since the mosaicing of split document images requires local information rather than global information. Hence, in this work, we explore a novel idea to obtain local values of the split documents by applying FT for smaller sized split documents of split document images. The proposed method assumes that the overlapping region is present at the right end of split image 1 and the left end of split image 2. The overlapping region is a common region, which helps in mosaicing.

Image Mosaicing is useful in a variety of vision and computer graphics applications like 3d vision, photogrammetry, satellite imagery, video images etc. The basic reason for Image Mosaicing is that camera vision is limited to 50 by 35 degrees, and Human Vision is limited to 200 degrees. By using Image Mosaicing we can ideally have 360 by 180 degrees vision. The modified version of Intensity Based Image Mosaicing algorithm has been proposed. Through this algorithm seams that appear while stitching images are not eliminated. This algorithm works both for single viewpoint as well as multiple viewpoints. The implementation aspect of the algorithm is discussed along with test results. The modified version algorithm for Image Mosaicing using Euclidean Warp and Bilinear Interpolation has been used for the purpose of comparison. We have used our own image data sets for experimenting.

Classic mosaic is one of the oldest and most durable art forms. There has been a growing interest in simulating classic mosaics from digital images recently. To be visually pleasing, a mosaic should satisfy the following constraints: tiles should be non-overlapping, tiles should align to the perceptually important edges in the underlying digital image, and orientation of the neighbouring tiles should vary smoothly across the mosaic. Most of the existing approaches operate in two steps: first they generate tile orientation field and then pack the tiles according to this field. However, previous methods perform these two steps based on heuristics or local optimisation which, in some cases, is not guaranteed to converge. Some other major disadvantages of previous approaches are: (i) either substantial user interaction or hard decision making such as edge detection is required before mosaicing starts (ii) the number of tiles per mosaic must be fixed beforehand, which may cause either undesired overlap or gap space between the tiles. In this work, we propose a novel approach by formulating the mosaic simulating problem in a global energy optimisation framework. Our algorithm also follows the two-step approach, but each step is performed with global optimisation. For the first step, we observe that the tile orientation constraints can be naturally formulated in an energy function that can be optimised with the α-expansion algorithm. For the second step of tightly packing the tiles, we develop a novel graph cuts based algorithm. Our approach does not require user interaction, explicit edge detection, or fixing the number of tiles, while producing results that are visually pleasing.

Classic mosaic is one of the oldest and most durable art forms. There has been a growing interest in simulating classic mosaics from digital images recently. To be visually pleasing, a mosaic should satisfy the following constraints: tiles should be non-overlapping, tiles should align to the perceptually important edges in the underlying digital image, and orientation of the neighbouring tiles should vary smoothly across the mosaic. Most of the existing approaches operate in two steps: first they generate tile orientation field and then pack the tiles according to this field. However, previous methods perform these two steps based on heuristics or local optimisation which, in some cases, is not guaranteed to converge. Some other major disadvantages of previous approaches are: (i) either substantial user interaction or hard decision making such as edge detection is required before mosaicing starts (ii) the number of tiles per mosaic must be fixed beforehand, which may cause either undesired overlap or gap space between the tiles. In this work, we propose a novel approach by formulating the mosaic simulating problem in a global energy optimisation framework. Our algorithm also follows the two-step approach, but each step is performed with global optimisation. For the first step, we observe that the tile orientation constraints can be naturally formulated in an energy function that can be optimised with the α-expansion algorithm. For the second step of tightly packing the tiles, we develop a novel graph cuts based algorithm. Our approach does not require user interaction, explicit edge detection, or fixing the number of tiles, while producing results that are visually pleasing.

Classic mosaic is one of the oldest and most durable art forms. There has been a growing interest in simulating classic mosaics from digital images recently. To be visually pleasing, a mosaic should satisfy the following constraints: tiles should be non-overlapping, tiles should align to the perceptually important edges in the underlying digital image, and orientation of the neighbouring tiles should vary smoothly across the mosaic. Most of the existing approaches operate in two steps: first they generate tile orientation field and then pack the tiles according to this field. However, previous methods perform these two steps based on heuristics or local optimisation which, in some cases, is not guaranteed to converge. Some other major disadvantages of previous approaches are: (i) either substantial user interaction or hard decision making such as edge detection is required before mosaicing starts (ii) the number of tiles per mosaic must be fixed beforehand, which may cause either undesired overlap or gap space between the tiles. In this work, we propose a novel approach by formulating the mosaic simulating problem in a global energy optimisation framework. Our algorithm also follows the two-step approach, but each step is performed with global optimisation. For the first step, we observe that the tile orientation constraints can be naturally formulated in an energy function that can be optimised with the α-expansion algorithm. For the second step of tightly packing the tiles, we develop a novel graph cuts based algorithm. Our approach does not require user interaction, explicit edge detection, or fixing the number of tiles, while producing results that are visually pleasing.

Image Stitching, also be called photo stitching, is the process of combining multiple photographic images with overlapping fields of view to produce a segmented panorama or high-resolution image. Image stitching is challenging in two fields. First, the sequenced photos taken from various angles will have different brightness. This will certainly lead to a un-nature stitched result with no harmony of brightness. Second, ghosting artifact due to the moving objects is also a common problem and the elimination of it is not an easy task. This paper presents several novel techniques that make the process of addressing the two difficulties significantly less labor-intensive while also efficient. For the brightness problem, each input image is blended by several images with different brightness. For the ghosting problem, we propose an intuitive technique according to a stitching line based on a novel energy map which is essentially a combination of gradient map which indicates the presence of structures and prominence map which determines the attractiveness of a region. The stitching line can easily skirt around the moving objects or salient parts based on the philosophy that human eyes mostly notice only the salient features of an image. We compare result of our method to those of 4 state-of-the-art image stitching methods and it turns out that our method outperforms the 4 methods in removing ghosting artifacts.

We define a multiresolution spline technique for combining two or more images into a larger image mosaic. In this procedure, the images to be splined are first decomposed into a set of band-pass filtered component images. Next, the component images in each spatial frequency band are assembled into a corresponding band-pass mosaic. In this step, component images are joined using a weighted average within a transition zone which is proportional in size to the wave lengths represented in the band. Finally, these band-pass mosaic images are summed to obtain the desired image mosaic. In this way, the spline is matched to the scale of features within the images themselves. When coarse features occur near borders, these are blended gradually over a relatively large distance without blurring or otherwise degrading finer image details in the neighborhood of the border.

We present an algorithm capable of making in real time image mosaics with enlarged field-of-view from the endoscopic video data stream. The algorithm is based on the method of Kourogi et al. (1999) which we extend to the case of endoscopic masks. The algorithm automatically finds the optimal affine transform between video frames and builds the enlarged field-of-view as an intervention-free side task. We apply our algorithm to endoscopic video sequences and compare it to the well-known image-mosaicing algorithm of Szeliski (1994). Our method turns out to be more robust, more than 3 times faster, having at the same time a 4 times smaller average motion estimation error: 0.19 pixel instead of 0.72 pixel between successive frames.

A dual-band infrared camera system based on a dual-band quantum well infrared photodetector (QWIP) has been developed for acquiring images from both the mid-wavelength (MWIR) and long-wavelength (LWIR) infrared spectral band. The system delivers exactly pixel-registered simultaneously acquired images. It has the advantage that appropriate signal and image processing permit to exploit differences in the characteristics of those bands. Thus, the camera reveals more information than a single-band camera. It helps distinguishing between targets and decoys and has the ability to defeat many IR countermeasures such as smoke, camouflage and flares. Furthermore, the system permits to identify materials (e.g. glass, asphalt, slate, etc.), to distinguish sun reflections from hot objects and to visualize hot exhaust gases. Furthermore, dedicated software for processing and exploitation in real-time extends the application domain of the camera system. One component corrects the images and allows for overlays with complementary colors such that differences become apparent. Another software component aims at a robust estimation of transformation parameters of consecutive images in the image stream for image registration purposes. This feature stabilizes the images also under rugged conditions and it allows for the automatic stitching of the image stream to construct large mosaic images. Mosaic images facilitate the inspection of large objects and scenarios and create a better overview for human observers. In addition, image based MTI (moving target indication) also for the case of a moving camera is under development. This component aims at surveillance applications and could also be used for camouflage assessment of moving targets.

Resumo O leque do Negro é um sistema deposicional aluvial que vem sendo construído pelo rio Negro na borda sudeste do Pantanal desde o Pleistoceno. Sua superfície apresenta feições geomorfológicas e drenagem atuais e reliquiares. O rio Negro flui de leste para oeste e integra a Bacia Hidrográfica do Alto Paraguai (BAP), sendo afluente da margem esquerda do rio Paraguai. A bacia de drenagem do rio Negro é um grande anfiteatro de erosão localizado no planalto de Maracaju-Campo Grande, esculpido em rochas paleozóicas e mesozóicas da Bacia do Paraná e em rochas neoproterozóicas do Grupo Cuiabá. O contato entre o planalto e a planície é marcado por escarpa retilínea de direção NNE. O rio Negro apresenta, no planalto, comportamento obsequente em relação ao mergulho das camadas da Bacia do Paraná. A evolução geomorfológica do leque do Negro teve início com a construção de um lobo antigo na saída do planalto, de provável idade pleistocênica, seguido de avulsão para SE e formação do lobo pré-atual, que termina na vazante Santa Clara. O sítio deposicional holocênico é representado pelo lobo atual, formado sobre depósitos distais do megaleque do Taquari, e por uma planície de meandros embutida em vale inciso na porção superior do leque. Terrenos com inúmeras lagoas, semelhantes às da Nhecolândia, ocorrem entre os lobos, constituindo formas relictas preservadas dentro da área do leque do rio Negro.

Introduction: Central pit craters are an unusual class of impact crater seen most commonly on the icy Galilean satellites and Mars. They are characterized by terraced rims and flattened floors with a pit at or near the center (see Fig 1). Their association with ice-rich surfaces has led most hypotheses for the origins of central pits to rely on the volatility or weakness of water ice [e.g. 1, 2, 3, 4]. Knowledge of pit dimensions is necessary to reveal the depth of pit influence, and can also be used to infer the quantities of material involved in the formation process. Such information will have important implications for the development and testing of pit formation theories; for example, variation of pit volume in craters formed on different terrain types may indicate a material property dependence of the pit formation mechanism. We will present topographic profiles of central pit craters on Ganymede. Measurements from these data are used to construct scaling trends to be used in future testing of numerical models.

Introduction: The icy Galilean moons of Europa, Ganymede and Callisto display exotic crater morphologies with no obvious analogue to craters on silicate bodies. As the Moon and Galilean satellites have similar gravity, differences in crater morphology are likely due to icy lithospheres being mechanically distinct from rocky bodies. The presence of subsurface liquid layers is also thought to affect crater morphology [1]. As the cratering process is affected by target properties, the study of crater morphology on the icy satellites provides a means for investigating the upper-crustal structure of these bodies. To understand the effects of layering on crater morphology, the underlying impact process in ice must first be understood. To understand crater formation requires two major elements: sufficient observational data, to inspire formation theories and provide ground truth data, and a means to test these proposed formation processesnumerical models. Craters on Europa are likely to be heavily affected by its sub-surface ocean as it is relatively close to the surface [2]. As Ganymede's ocean is at a greater depth, craters on Ganymede provide better observational data for the investigation of impact into unlayered ice. We present scaling trends of complex crater dimensions drawn from topographic profiles of craters on Ganymede. We compare these trends with those of analogous features in craters on the Moon, and investigate the relationship between these morphological trends and target strength using dynamic modeling. We collected topographic profiles of 48 craters on both dark and bright terrains of Ganymede from Galileo data. Most craters profiled were relatively young so that good comparison could be made with fresh impact craters on the moon and final craters produced by our computer models. Several cross-sectional profiles were taken of each impact crater so that any artifacts introduced as a result of the profiling technique, or features superimposed after impact, could be identified and removed. We then collected measurements from each crater, including crater depths and diameters, heights and widths of central features, and slope angles. When a variance in (e.g.) central peak width between each of the topographic profiles of the same crater was evident, the maximum value was adopted. The variety of scaling trends presented here represent the variation in maximum values.

The Global Rain Forest Mapping (GRFM) project is an international endeavour led by the National Space Development Agency of Japan (NASDA), with the aim of producing spatially and temporally contiguous Synthetic Aperture Radar (SAR) data sets over the tropical belt on the Earth by use of the JERS-1 L-band SAR, through the generation of semi-continental, 100 m resolution, image mosaics. The GRFM project relies on extensive collaboration with the National Aeronautics and Space Administration (NASA), the Joint Research Centre of the European Commission (JRC) and the Japanese Ministry of International Trade and Industry (MITI) for data acquisition, processing, validation and product generation. A science programme is underway in parallel with product generation. This involves the agencies mentioned above, as well as a large number of international organizations, universities and individuals to perform eld activities and data analysis at di erent levels. The GRFM project was initiated in 1995 and, through a dedicated data acquisition policy by NASDA, data acquisitions could be completed within a 1.5-year period, resulting in a spatially and temporally homogeneous coverage to encompass the entire Amazon Basin from the Atlantic to the Paci c; Central America up to the Yucatan Peninsular in Mexico; equatorial Africa from Madagascar and Kenya in the east to Sierra Leone in the west; and southeast Asia, including Papua New Guinea and northern Australia. Over the Amazon and Congo river basins, the project aimed to provide complete cover at two di erent seasons, featuring the basins at high and low water. In total, the GRFM acquisitions comprise some 13 000 SAR scenes, which are currently in the course of being processed and compiled into image mosaics. In March 1999, SAR mosaics over the Amazon Basin (one out of two seasonal coverages) and equatorial Africa (both seasonal coverages) were completed; the data are available on CD-ROM and, at a coarser resolution, via the Internet. Coverage of the second-season Amazon and Central America will be completed during 1999, with the southeast Asian data sets following thereafter. All data are being provided free of charge to the international science community for research and educational purposes.

We introduce a mobile system for high resolution panoramic image creation. In our system the user can rotate the camera arbitrarily and see the preview panorama in real-time. The system automatically captures high resolution images and generates a high-quality result with unlimited viewing angle. We employ a coarse-to-fine method for high-quality registration, and a seam-finding method to remove the ghosting effect due to moving objects. The proposed system has been tested on several types of camera phones, and the tests reveal that the system can efficiently provide a high-quality panoramic image in spite of the low computational power and memory available in such devices.

Le travail présenté dans cette thèse a été mené au sein de l'équipe Scènes Dynamiques du Laboratoire d'Automatique I 3 D 1 de l'Université des Sciences et Technologies de Lille 1. Je tiens à remercier Monsieur Olivier COLOT, Professeur à l'USTL, pour m'avoir fait l'honneur de présider le jury de cette thèse. Je tiens à exprimer toute ma gratitude à Monsieur le Professeur Jack-Gérard POSTAIRE, mon Co-Directeur de recherche, pour l'intérêt qu'il a apporté à mes travaux, pour la confiance qu'il m'a accordée. Ses conseils, ses constants encouragements et sa grande disponibilité m'ont permis de mener à bien ce travail. Ma reconnaissance et mes vifs remerciements vont à Monsieur Yassine RUICHEK, mon second Co-Directeur de recherche, Maître de Conférences à l'Université de Technologie de Belfort-Montbéliard. Ses conseils, sa rigueur scientifique ainsi que ses idées originales et novatrices m'ont été très précieux pour mener à bien ces années de recherche. Je suis très honoré par la présence dans ce jury de Monsieur Abderrafiaa KOUKAM,

Text extraction is one of the key tasks in document image analysis. Automatic text extraction without characters recognition capabilities is to extract regions just contains text. The text extraction process includes detection, localization, segmentation and enhancement of the text from the given input image. In this paper we present a comparative study and performance evaluation of various text extraction techniques.