Publications

In this paper we introduce a novel surveillance system, which uses 3D information extracted from multiple cameras to detect, track and re-identify people. The detection method is based on a 3D Marked Point Process model using two pixel-level features extracted from multi-plane projections of binary foreground masks, and uses a stochastic optimization framework to estimate the position and the height of each person. We apply a rule based Kalman-filter tracking on the detection results to find the object-to-object correspondence between consecutive time steps. Finally, a 3D body model based long-term tracking module connects broken tracks and is also used to re-identify people

AD-HOC (Appearance Driven Human tracking with Occlusion Classification) is a complete framework for multiple people tracking in video surveillance applications in presence of large occlusions. The appearance-based approach allows the estimation of the pixel-wise shape of each tracked person even during the occlusion. This peculiarity can be very useful for higher level processes, such as action recognition or event detection. A first step predicts the position of all the objects in the new frame while a MAP framework provides a solution for best placement. A second step associates each candidate foreground pixel to an object according to mutual object position and color similarity. A novel definition of non-visible regions accounts for the parts of the objects that are not detected in the current frame, classifying them as dynamic, scene or apparent occlusions. Results on surveillance videos are reported, using in-house produced videos and the PETS2006 test set.

Several industrial applications do require embedded solutionsboth for compacting the hardware occupation and reducing energy consumption, and for achieving high speed performance. This paper presents a computer vision system developed for correcting image skew in applications for banknote analysis and classification. The system must be very efficient and run on a fixed-point DSP with limited computational resources. Consequently, we propose three innovative improvements to basic and general-purpose image processing techniques that can be helpful in other computer vision applications on embedded devices. In particular, we address: a) an efficient labeling with an unionfind approach for hole filling, b) a fast Hough transform implementation, and c) a very high-speed estimation of affinetransformation for skew correction. The reported results demonstrate both the accuracy and the efficiency of the system,also in presence of severe skew. In terms of efficiency, the computational time is reduced of about two orders of magnitude.

Video surveillance is becoming the technology of choice for monitoring crowded areas for security threats. While video provides ample information for human inspectors, there is a great need for robust automated techniques that can efficiently detect anomalous behavior in streaming video from single ormultiple cameras. In this work we synergistically combine two state-of-the-art methodologies. The rst is the ability to track and label single person trajectories in a crowded area using multiple video cameras, and the second is a new class of novelty detection algorithms based on spectral analysis of graphs. By representing the trajectories as sequences of transitions betweennodes in a graph, shared individual trajectories capture only a small subspace of the possible trajectories on the graph. This subspace is characterized by large connected components of the graph, which are spanned by the eigenvectors with the low eigenvalues of the graph Laplacian matrix. Using this technique, we develop robust invariant distance measures for detectinganomalous trajectories, and demonstrate their application on realvideo data.

The interest of the research community in creating reference datasets for performance analysis is always very high. Although new datasets, collecting large amounts of video footage are spreading in surveillance and forensics, few bench-marks with annotation data are available for testing specific tasks and especially for 3D/multi-view analysis. In this paper we present 3DPeS, a new dataset for 3D/multi- view surveillance and forensic applications. This has been designed for discussing and evaluating research results in people re-identification and other related activities (people detection, people segmentation and people tracking). The new assessed version of the dataset contains hundreds of video sequences of 200 people taken from a multi-camera distributed surveillance system over several days, with different light conditions; each person is detected multiple times and from different points of view. In surveillance scenarios, the dataset can be exploited to evaluate people reacquisition, 3D body models and people activity reconstruction algorithms. In forensics it can be adopted too, by relaxing some constraints (e.g. real time) and neglecting some information (e.g. calibration). Some results on this new dataset are presented using state of the art methods for people re-identification as a benchmark for future comparisons.

People trajectory analysis is a recurrent task inmany pattern recognition applications, such as surveillance,behavior analysis, video annotation, and many others. In thispaper we propose a new framework for analyzing trajectoryshape, invariant to spatial shifts of the people motion in thescene. In order to cope with the noise and the uncertainty ofthe trajectory samples, we propose to describe the trajectoriesas a sequence of angles modelled by distributions of circularstatistics, i.e. a mixture of von Mises (MovM) distributions.To deal with MovM, we define a new specific EM algorithmfor estimating the parameters and derive a closed form of theBhattacharyya distance between single vM pdfs. Trajectories arethen modelled with a sequence of symbols, corresponding tothe most suitable distribution in the mixture, and comparedeach other after a global alignment procedure to cope withtrajectories of different lengths. The trajectories in the trainingset are clustered according with their shape similarity in an offlinephase, and testing trajectories are then classified with aspecific on-line EM, based on sufficient statistics. The approachis particularly suitable for classifying people trajectories in videosurveillance, searching for abnormal (i.e. infrequent) paths. Testson synthetic and real data are provided with also a completecomparison with other circular statistical and alignment methods.

This paper presents a method for recognising human actions bytracking body parts without using artificial markers. A sophisticated appearance-based tracking able to cope with occlusions is exploited to extract a probability map for each moving object. A segmentation technique based on mixture of Gaussians (MoG) is then employed to extract and track significantpoints on this map, corresponding to significant regions on the human silhouette. The evolution of the mixture in time is analysed by transforming it in a sequence of symbols (corresponding to a MoG). The similarity between actions is computed by applying global alignment and dynamic programming techniques to the corresponding sequences and using a variational approximation of the Kullback-Leibler divergence to measure the dissimilarity between two MoGs. Experiments on publicly available datasets and comparison with existing methods are provided.

This paper provides an analysis on relevance feedback techniques in a multimedia system designed for the interactive exploration and annotation of artistic collections, in particular illuminated manuscripts. The relevance feedback is presented not only as a very effective technique to improve the performance of the system, but also as a clever way to increase the user experience, mixing the interactive surfing through the artistic content with the possibility to gather valuable information from the user, and consequently improving his retrieval satisfaction. We compare a modification of the Mean-Shift Feature Space Warping algorithm, as representative of the standard RF procedures, and a learning-based technique based on transduction, considered in order to overcome some limitation of the previous technique. Experiments are reported regarding the adopted visual features based on covariance matrices.