Abstract: Many video sequences consist of a locally dynamic background containing moving foreground subjects. In this paper we propose a novel way of re-displaying these sequences, by giving the user control over a virtual camera frame. Based on video mosaicing, we first compute a static high quality background panorama. After segmenting and removing the foreground subjects from the original video, the remaining elements are merged into a dynamic background panorama, which seamlessly extends the original video footage. We then re-display this augmented video by warping and cropping the panorama. The virtual camera can have an enlarged field-of-view and a controlled camera motion. Our technique is able to process videos with complex camera motions, reconstructing high quality panoramas without parallax artefacts, visible seams or blurring, while retaining repetitive dynamic elements.

View: [PDF]

Abstract: We present a novel method for 3D shape acquisition, based on mobile structured light. Unlike classical structured light methods, in which a static projector illuminates the scene with dynamic illumination patterns, mobile structured light employs a moving projector translated at a constant velocity in the direction of the projector's horizontal axis, emitting static or dynamic illumination. For our approach, a time multiplexed mix of two signals is used: (1) a wave pattern, enabling the recovery of point-projector distances for each point observed by the camera, and (2) a 2D De Bruijn pattern, used to uniquely encode a sparse subset of projector pixels. Based on this information, retrieved on a per (camera) pixel basis, we are able to estimate a sparse reconstruction of the scene. As this sparse set of 2D-3D camera-scene correspondences is sufficient to recover the camera location and orientation within the scene, we are able to convert the dense set of point-projector distances into a dense set of camera depths, effectively providing us with a dense reconstruction of the observed scene. We have verified our technique using both synthetic and real-world data. Our experiments display the same level of robustness as previous mobile structured light methods, combined with the ability to accurately estimate dense scene structure and accurate camera/projector motion without the need for prior calibration.

View: [PDF] [YouTube]

Abstract: In this paper we present a novel decolorization strategy, based on image fusion principles. We show that by defining proper inputs and weight maps, our fusion-based strategy can yield accurate decolorized images, in which the original discriminability and appearance of the color images are well preserved. Aside from the independent R,G,B channels, we also employ an additional input channel that conserves color contrast, based on the Helmholtz-Kohlrausch effect. We use three different weight maps in order to control saliency, exposure and saturation. In order to prevent potential artifacts that could be introduced by applying the weight maps in a per pixel fashion, our algorithm is designed as a multi-scale approach. The potential of the new operator has been tested on a large dataset of both natural and synthetic images. We demonstrate the effectiveness of our technique, based on an extensive evaluation against the state-of-the-art grayscale methods, and its ability to decolorize videos in a consistent manner.

Abstract: In this paper we introduce a novel approach to restoring a single image degraded by atmospheric phenomena such as fog or haze. The presented algorithm allows for fast identification of hazy regions of an image, without making use of expensive optimization and refinement procedures. By applying a single per pixel operation on the original image, we produce a 'semi-inverse' of the image. Based on the hue disparity between the original image and its semi-inverse, we are then able to identify hazy regions on a per pixel basis. This enables for a simple estimation of the airlight constant and the transmission map. Our approach is based on an extensive study on a large data set of images, and validated based on a metric that measures the contrast but also the structural changes. The algorithm is straightforward and performs faster than existing strategies while yielding comparative and even better results. We also provide a comparative evaluation against other recent single image dehazing methods, demonstrating the efficiency and utility of our approach.

Abstract: In this paper we present a novel method for 3D structure acquisition, based on structured light. Unlike classical structured light methods, in which a static projector illuminates a scene with time-varying illumination patterns, our technique makes use of a moving projector emitting a static striped illumination pattern. This projector is translated at a constant velocity, in the direction of the projector's horizontal axis. Illuminating the object in this manner allows us to perform a per pixel analysis, in which we decompose the recorded illumination sequence into a corresponding set of frequency components. The dominant frequency in this set can be directly converted into a corresponding depth value. This per pixel analysis allows us to preserve sharp edges in the depth image. Unlike classical structured light methods, the quality of our results is not limited by projector or camera resolution, but is solely dependent on the temporal sampling density of the captured image sequence. Additional benefits include a significant robustness against common problems encountered with structured light methods, such as occlusions, specular reflections, subsurface scattering, interreflections, and to a certain extent projector defocus.

View: [PDF] [YouTube]

Abstract: In this paper we present a method for efficient calibration of a screen-camera setup, in which the camera is not directly facing the screen. A spherical mirror is used to make the screen visible to the camera. Using Gray code illumination patterns, we can uniquely identify the reflection of each screen pixel on the imaged spherical mirror. This allows us to compute a large set of 2D-3D correspondences, using only two sphere locations. Compared to previous work, this means we require less manual interventions, combined with a more robust screen pixel detection scheme. This results in a consistent improvement in accuracy, which we illustrate with experiments on both synthetic and real data.

View: [PDF]

Abstract: In this paper, our objective is to facilitate the creation of novel human poses by synthesizing images. Existing approaches commonly deform one single image, which often results in a distorted image due to texture and illumination artefacts. We present a novel image-based pose synthesis technique that accurately reconstructs texture details by combining information from multiple photographs. Given a user-specified 2D target pose, our solution merges different parts of the input photographs in order to conform to the desired pose, solely using 2D operations. We illustrate how novel poses can be generated from only a few example images, requiring little user intervention.

View: [PDF]

Abstract: We propose an efficient technique for normal map acquisition, using a cheap and easy to build setup. Our setup consists solely of off-the-shelf components, such as an LCD screen, a digital camera and a linear polarizer filter. The LCD screen is employed as a linearly polarized light source emitting gradient patterns, whereas the digital camera is used to capture the incident illumination reflected off the scanned object's surface. Also, by exploiting the fact that light emitted by an LCD screen has the property of being linearly polarized, we use the filter to surpress any specular highlights. Based on the observed Lambertian reflection of only four different light patterns, we are able to obtain a detailed normal map of the scanned surface. Overall, our techniques produces convincing results, even on weak specular materials.

View: [PDF]

Abstract: This work addresses the practical problem of keeping a camera network calibrated during a recording session. When dealing with real-time applications, a robust calibration of the camera network needs to be assured, without the burden of a full system recalibration at every (un)intended camera displacement. In this paper we present an efficient algorithm to detect when the extrinsic parameters of a camera are no longer valid, and reintegrate the displaced camera into the previously calibrated camera network. When the intrinsic parameters of the cameras are known, the algorithm can also be used to build ad-hoc distributed camera networks, starting from three calibrated cameras. Recalibration is done using pairs of essential matrices, based on image point correspondences. Unlike other approaches, we do not explicitly compute any 3D structure for our calibration purposes.

View: [PDF]

Abstract: Recent developments in the consumer market have indicated that the average user of a personal computer is likely to also own a webcam. With the emergence of this new user group will come a new set of applications, which will require a user-friendly way to calibrate the position of the camera compared to the location of the screen. This paper presents a fully automatic method to calibrate a screen-camera setup, using a single moving spherical mirror. Unlike other methods, our algorithm needs no user intervention other then moving around a spherical mirror. In addition, if the user provides the algorithm with the exact radius of the sphere in millimeters, the scale of the computed solution is uniquely defined.

View: [PDF]

Abstract: We present a camera-projection system for single- or multiparty telepresence which allows for correct eye gaze, and unlike standard videophony, provides a great deal of spatial context. The unique feature of our system is a combination of omnidirectional video capture and display from corresponding projection centers. In essence, this creates a virtual overlap between the screen and the camera, which results in the participant looking directly into the camera whilst looking at the display. The novelty in this approach is that we correct eye gaze without the need of interpolating multiple views. With a given external calibration of the camera-projector setup, the mapping used for this system has to be calculated only once. This makes it possible for the algorithm to be implemented on commodity hardware and the GPU.

View: [PDF (SVR)] [Poster (ProCams 2006)]

Abstract (partial): In this dissertation we introduce a new set of 3D shape acquisition methods, which we have called mobile structured light. Unlike the more classical structured light methods, in which a static projector illuminates a (static) scene with a variety of time-varying illumination patterns, our proposed technique makes use of a mobile projector emitting a single static sinusoidal illumination pattern.

View: [PDF]

Abstract: We provide the reading with three subjects: (1) an overview of the state of the art in the NPR domain, anno 2005, (2) a detailed theoretical background needed for painterly rendering of images and videos, (3) details about our implementation and screenshots of the results.

View: [PDF]