We develop a powerful probabilistic framework for the local characterization of surfaces and edges in range images. We use the geometrical nature of the data to derive an analytic expression for the joint probability density function (pdf) for the random variables used to model the ranges of a set of pixels in a local neighborhood of an image. We decompose this joint pdf by considering independently the cases where two real world points corresponding to two neighboring pixels are locally on the same real world surface or not. In particular, we show that this joint pdf is linked to the Voigt pdf and not to the Gaussian pdf as it is assumed in some applications. We apply our framework to edge detection and develop a locally adaptive algorithm that is based on a probabilistic decision rule. We show in an objective evaluation that this new edge detector performs better than prior art edge detectors. This proves the benefits of the probabilistic characterization of the local neighborhood as a tool to improve applications that involve range images.

Minimally conscious state (MCS) is a neurological syndrome in which the patient shows signs of partial consciousness after having emerged from unresponsive wakefulness syndrome (UWS), which itself follows a state of coma. Distinguishing between MCS and UWS is complex and has major impact on the clinical management and prognosis of affected patients. Research on disorders of consciousness (DoC) has revealed that (1) visual pursuit, i.e. the ability of a patient to track a moving stimulus, is one of the most decisive clinical signs for establishing the MCS/UWS distinction, and that (2) the most effective moving stimulus for visual pursuit assessment is a mirror where the patient can see his/her own face. In clinical practice, while this guidance is widely followed, the visual pursuit ability is typically assessed on the basis of the clinician's opinion only, i.e. in a subjective thus biased manner. In this paper, we present a new system using cameras and computer vision techniques, which helps clinicians to objectify the assessment of visual pursuit. Our system is specifically designed to work with the moving mirror stimulus in order to follow the recommended, well-established clinical setup. We validate our system on healthy control subjects and give preliminary results obtained with DoC patients.

Drowsiness is a major cause of accidents, in particular in road transportation. It is thus crucial to develop robust drowsiness monitoring systems. There is a widespread agreement that the best way to monitor drowsiness is by closely monitoring symptoms of drowsiness that are directly linked to the physiology of an operator such as a driver. The best systems are completely transparent to the operator until the moment he/she must react. In transportation, cameras placed in the passenger compartment and looking at least at the face of the driver are most likely the best way to sense physiology related symptoms such as facial expressions and the fine behavior of the eyeballs and eyelids. We present here the new database called DROZY that provides multiple modalities of data to tackle the design of drowsiness monitoring systems and related experiments. We also present two novel systems developed using this database that can make predictions about the speed of reaction of an operator by using near-infrared intensity and range images of his/her face.

We present a new method to estimate the clutter-plus-noise covariance matrix used to compute an adaptive filter in space-time adaptive processing (STAP). The method computes a ML estimate of the clutter scattering coefficients using a Bayesian framework and knowledge on the structure of the covariance matrix. A priori information on the clutter statistics is used to regularize the estimation method. Other estimation methods based on the computation of the power spectrum using for instance the periodogram are compared to our method. The result in terms of SINR loss shows that the proposed method outperforms the other ones.

With the advent of new, more widely available very high resolution (VHR) SAR and optical sensor satellite constellations, the issue of jointly exploiting the corresponding imaging data in the best possible way naturally arises. Our ultimate goal is the pixel-level fusion of these modalities and their visualization in a 3D context, i.e. draped over terrain data, itself obtained from interferometric SAR processing. In this paper, we investigate the issues associated with adapting tools originally developed for low-resolution C-band SAR imagery to high-resolution X-band SAR imagery (both spaceborne). Specifically, we consider the issues associated with SAR interferometry (InSAR) and more particularly those associated with complex interpolation and phase unwrapping. We found that the existing chirp-Z transform complex interpolator previously used at low resolution is perfectly suitable for handling VHR SAR data. We also found that, while the selected phase unwrapping procedure works well when the resolution is reduced by averaging, it is deficient at full resolution, in part due micro-reliefs and specific phase responses.

This paper presents a new approach for tracking objects in complex situations such as people in a crowd or players on a soccer field. Each object in the image is represented by several interest points (IPs). These IPs are obtained using a color version of the Harris IP detector. Each IP is characterized by the local appearance (chromatic first-order local jet) of the object and by geometric parameters. We track objects by matching IPs from image to image based on the Mahalanobis distance. The approach is robust to occlusion. Performance is illustrated by some examples.

This article presents a modular architecture for multicamera tracking in the context of sports broadcasting. For each video stream, a geometrical module continuously performs the image-to-model homography estimation. A local-feature based tracking module tracks the players in each view. A supervisor module collects, associates and fuses the data provided by the tracking modules. The originality of the proposed system is three-fold. First, it allows to localize the targets on the ground with rotating and zooming cameras; second, it does not use background modeling techniques; and third, the local tracking can cope with severe occlusions. We present experimental results on raw TV-camera footage of a soccer game.

This article focuses on the field-to-image registration for TV sports-related image processing through basic geometrical reasoning. We propose a robust automatic line-based method that can cope with imprecisions in the line detection steps through a hypothesis generation/pruning paradigm where the number of hypotheses is reduced by using the vanishing points. The results are satisfying and illustrated with examples from sport.

We describe an important element of an automatic sport analysis system. This element continuously estimates the image-to-model homography from the video stream of a single camera. Here, we focus on the incremental image-to-image updating of the homography matrix, which greatly facilitates real-time operation. This updating relies on the automatic tracking of interest lines and points and on their use for robust homography estimation. Results on real video sequences are shown. 1

We address the problem of detecting slow-moving targets using a non-sidelooking monostatic space-time adaptive processing (STAP) radar. The construction of optimum weights at each range implies the estimation of the clutter covariance matrix. This is typically done by straight averaging of neighboring data snapshots. The range-dependence of these snapshots generally results in poor performance. We present two new methods that handle the range-dependence by exploiting the geometry of the direction-Doppler curves.

We describe an experimental, model-based automatic target recognition (ATR) system, called XTRS, for recognizing tactical vehicles in real or synthetic laser-radar (LADAR) range and intensity images corresponding to a forwardlooking, CO/sub 2/ laser radar (LADAR) that is carried either on a ground vehicle or on an airborne platform. Various aspects of the system's operation are illustrated through a variety of examples. Generic techniques are highlighted whenever possible. A first such technique is the use of feature-indicating interest images to focus attention on specific areas of the input imagery. A second is the use of an application-independent matching engine for matching features extracted from the imagery against an application-dependent appearance model hierarchy that represents the objects to be recognized. A third generic technique is the system's architectures and its control mechanism. Following the description of XTRS, we discuss XTRS's recognition performance on real data collected with the groundbased version of the ladar sensor. We then provide a detailed account of XTRS's performance on synthetic datasets created to rest the limits of system performance. Finally, we briefly discuss the use of XTRS in conjunction with the airborne version of the sensor. Overall, more than 1500 range and intensity image pairs were used throughout XTRS's development.

The application of adaptive (i.e., data-dependent) mathematical morphology techniques to range imagery, i.e., the use of structuring elements (SEs) that automatically adjust to the gray-scale values in a range image in order to deal with features of known physical sizes, is discussed. The technique is applicable to any type of image for which the distance to a scene element is available for each pixel.

A new approach to shape matching and knowledge-based signal processing, called functional template correlation (FTC), is presented in the context of 3D object recognition. Incorporating aspects of fuzzy set theory, functional templates (FTs) define, for each template point, arbitrarily complex tolerances of input image values. With this approach, object- and sensor-dependent knowledge is easily encoded in FTs, making it possible to effectively deal with uncertain or variable object appearance noise occlusion, and articulation. The output of FTC is a map of match scores, reflecting pixel-by-pixel belief of whether or not an encoded shape is present. Such maps can be combined using simple rules of arithmetic and are used to guide selective attention. The use of FTC is illustrated with examples from work in automatic recognition.>

A model-based technique is presented for recognizing 3-D objects using a novel object representation and a novel generic matching engine. Each object to be recognized is defined by an appearance model (AM) describing the expected appearances of the object over a range of aspects in a specific type of imagery. The AMs for all the objects to be recognized are organized in an AM hierarchy (AMH) defining object classes. This AMH, together with an image event extracted from the input imagery and primitives obtained by decomposing the event, constitute the input to the matching engine. This engine determines which object in the AMH most likely corresponds to the event and its primitives. The approach described is at the heart of an automatic target recognition system built by the authors.>

Although little known, mathematical morphology (MM) offers great potential in the areas of image enhancement, feature extraction, and object recognition. MM has the intrinsic ability to quantitatively analyze object shapes in both 2 and 3 dimensions. Using MM to extract features and recognize objects in range imagery seems particularly appropriate since range data is a natural source of shape information. We present several experimental results of applying MM techniques to real and synthetic range imagery, both for noise removal and feature extraction.

Range images obtained from infrared ranging sensors often provide detailed silhouettes of objects in the field of view. Elements of a system for extracting and analyzing such range silhouettes are proposed. The effectiveness of the system implemented so far is demonstrated on real data. The other elements of the ultimate system are also discussed, but, only at a conceptual level.

This paper presents a silhouette-slice theorem for convex opaque 3-D objects. The theorem states that the 2-D Curvature Transform (CT) of any silhouette contour is a slice of the 3-D CT of the object surface at some appropriate orientation. The 2-D and 3- D CT's are defined as curvature functions on the Gaussian circle and sphere of the silhouette and object, respectively. The new theorem and transforms are shown to be the counterparts in silhouette imaging of the projection-slice theorem and Fourier Transforms of line-integral projection imaging.