Several pattern recognition and classification techniques have been applied to the biometrics domain. Among them, an interesting technique is the Scale Invariant Feature Transform (SIFT), originally devised for object recognition. Even if SIFT features have emerged as a very powerful image descriptors, their employment in face analysis context has never been systematically investigated. This paper investigates the application of the SIFT approach in the context of face authentication. In order to determine the real potential and applicability of the method, different matching schemes are proposed and tested using the BANCA database and protocol, showing promising results. 1.

We address the problem of robust multi-target tracking within the application of hockey player tracking. Although there has been extensive work in multi-target tracking, there is no existing visual tracking system that can automatically and robustly track a variable number of targets and correctly maintain their identities with a monocular camera regardless of background clutter, camera motion and frequent mutual occlusion between targets. We build our system on the basis of the previous work by Okuma et al. [OTdF + 04]. The particle filter technique is adopted and modified to fit into the multi-target tracking framework. A rectification technique is employed to map the locations of players from the video frame coordinate system to the standard hockey rink coordinates so that the system can compensate for camera motion and the dynamics of players on the rink can be improved by a second order auto-regression model. A global nearest neighbor data association algorithm is introduced to assign boosting detections to the existing tracks for the proposal distribution in particle filters. The mean-shift algorithm is embedded into the particle filter framework to stabilize the trajectories of the targets for robust tracking during mutual occlusion. The color model of the targets is also improved by the kernel introduced by mean-shift. Experimental results show that our system is able to correctly track all the targets in the scene even if they are partially or completely occluded for a period of time.

Figure 1 On the left, a panoramic sequence of images showing vignetting artifacts. Note the change in brightness at the edge of each image. Although the effect is visually subtle, this brightness change corresponds to a 20 % drop in image intensity from the center to the corner of each image. On the right, the same sequence after vignetting is removed. We discuss calibration and removal of “vignetting ” (radial falloff) and exposure (gain) variations from sequences of images. Unique solutions for vignetting, exposure and scene radiances are possible when the response curve is known. When the response curve is unknown, an exponential ambiguity prevents us from recovering these parameters uniquely. However, the vignetting and exposure variations can nonetheless be removed from the images without resolving this ambiguity. Applications include panoramic image mosaics, photometry for material reconstruction, imagebased rendering, and preprocessing for correlation-based vision algorithms. 1.

We propose a considerably faster approximation of the well known SIFT method. The main idea is to use efficient data structures for both, the detector and the descriptor. The detection of interest regions is considerably speed-up by using an integral image for scale space computation. The descriptor which is based on orientation histograms, is accelerated by the use of an integral orientation histogram. We present an analysis of the computational costs comparing both parts of our approach to the conventional method. Extensive experiments show a speed-up by a factor of eight while the matching and repeatability performance is decreased only slightly.

Our goal is an automated 2D-image-pair registration algorithm capable of aligning images taken of a wide variety of natural and man-made scenes as well as many medical images. The algorithm should handle low overlap, substantial orientation and scale differences, large illumination variations, and physical changes in the scene. An important component of this is the ability to automatically reject pairs that have no overlap or have too many differences to be aligned well. We propose a complete algorithm including techniques for initialization, for estimating transformation parameters, and for automatically deciding if an estimate is correct. Keypoints extracted and matched between images are used to generate initial similarity transform estimates, each accurate over a small region. These initial estimates are rank-ordered and tested individually in succession. Each estimate is refined using the Dual-Bootstrap ICP algorithm, driven by matching of multiscale features. A three-part decision criteria, combining measurements of alignment accuracy, stability in the estimate, and consistency in the constraints, determines whether the refined transformation estimate is accepted as correct. Experimental results on a data set of 22 challenging image pairs show that the algorithm effectively aligns 19 of the 22 pairs and rejects 99.8 percent of the misalignments that occur when all possible pairs are tried. The algorithm substantially out-performs algorithms based on keypoint matching alone.

Abstract—Video stabilization is an important video enhancement technology which aims at removing annoying shaky motion from videos. We propose a practical and robust approach of video stabilization that produces full-frame stabilized videos with good visual quality. While most previous methods end up with producing smaller size stabilized videos, our completion method can produce fullframe videos by naturally filling in missing image parts by locally aligning image data of neighboring frames. To achieve this, motion inpainting is proposed to enforce spatial and temporal consistency of the completion in both static and dynamic image areas. In addition, image quality in the stabilized video is enhanced with a new practical deblurring algorithm. Instead of estimating point spread functions, our method transfers and interpolates sharper image pixels of neighboring frames to increase the sharpness of the frame. The proposed video completion and deblurring methods enabled us to develop a complete video stabilizer which can naturally keep the original image quality in the stabilized videos. The effectiveness of our method is confirmed by extensive experiments over a wide variety of videos. Index Terms—Video analysis, video stabilization, video completion, motion inpainting, sharpning and deblurring, video enhancement. æ 1

Figure 1: A cylindrical projection of an environment map constructed using Envisor with a camera on a tripod. One important component of modeling new scenes is the acqui-One of the main goals of anywhere augmentation is the develop- sition of an environment map. As an image-based representation ment of automatic algorithms for scene acquisition in augmented of the light distribution around a single position, environment maps reality systems. In this paper, we present Envisor, a system for have many uses in AR systems. Most commonly, they can be used online construction of environment maps in new locations. To ac- for realistic shading of virtual geometry [1, 8, 12] for more seamcomplish this, Envisor uses vision-based frame to frame and land- less integration of virtual objects into the physical scene. They are mark orientation tracking for long-term, drift-free registration. For also useful for remote presence applications [25], as a simple way additional robustness, a gyroscope / compass orientation unit can of representing a remote environment, e.g. as a backdrop in a teleoptionally be used for hybrid tracking. The tracked video is then collaboration system, or in low-bandwidth first-person interfaces projected into a cubemap frame by frame. Feedback is presented like QuickTime VR models [18]. to the user to help avoid gaps in the cubemap, while any remain-In this paper, we present Envisor, a system for the automatic, oning gaps are filled by texture diffusion. The resulting environment line construction of environment maps using a hand-held or headmap

With the addition of free programmable components to modern graphics hardware, graphics processing units (GPUs) become increasingly interesting for general purpose computations, especially due to utilizing parallel buffer processing. In this paper we present methods and techniques that take advantage of modern graphics hardware for real-time tracking and recognition of feature-points. The focus lies on the generation of feature vectors from input images in the various stages. For the generation of feature-vectors the Scale Invariant Feature Transform (SIFT) method [Low04a] is used due to its high stability against rotation, scale and lighting condition changes of the processed images. We present results of the various stages for feature vector generation of our GPU implementation and compare it to the CPU version of the SIFT algorithm. The approach works well on Geforce6 series graphics board and above and takes advantage of new hardware features, e.g. dynamic branching and multiple render targets (MRT) in the fragment processor [KF05]. With the presented methods feature-tracking with real time frame rates can be achieved on the GPU and meanwhile the CPU can be used for other tasks.

The task of registering video frames with a static model is a common problem in many computer vision domains. The standard approach to registration involves finding point correspondences between the video and the model and using those correspondences to numerically determine registration transforms. Current methods locate video-to-model point correspondences by assembling a set of reference images to represent the model and then detecting and matching invariant local image features between the video frames and the set of reference images. These methods work well when all video frames can be guaranteed to contain a sufficient number of distinctive visual features. However, as we demonstrate, these methods are prone to severe misregistration errors in domains where many video frames lack distinctive image features. To overcome these errors, we introduce a concept of local distinctiveness which allows us to find model matches for nearly all video features, regardless of their distinctiveness on a global scale. We present results from the American football domain—where many video frames lack distinctive image features—which show a drastic improvement in registration accuracy over current methods. In addition, we introduce a simple, empirical stability test that allows our method to be fully automated. Finally, we present a registration dataset from the American football domain we hope can be used as a benchmarking tool for registration methods. 1.

In this paper we propose to transform an image descriptor so that nearest neighbor (NN) search for correspondences becomes the optimal matching strategy under the assumption that inter-image deviations of corresponding descriptors have Gaussian distribution. The Euclidean NN in the transformed domain corresponds to the NN according to a truncated Mahalanobis metric in the original descriptor space. We provide theoretical justification for the proposed approach and show experimentally that the transformation allows a significant dimensionality reduction and improves matching performance of a state-of-the art SIFT descriptor. We observe consistent improvement in precision-recall and speed of fast matching in tree structures at the expense of little overhead for projecting the descriptors into transformed space. In the context of SIFT vs. transformed M-SIFT comparison, tree search structures are evaluated according to different criteria and query types. All search tree experiments confirm that transformed M-SIFT performs better than the original SIFT. 1.