Roberto Vezzani

Probabilistic posture classification for Human-behavior analysis

Computer vision and ubiquitous multimedia access nowadays make feasible the development of a mostly automated system for human-behavior analysis. In this context, our proposal is to analyze human behaviors by classifying the posture of the monitored person and, consequently, detecting corresponding events and alarm situations, like a fall. To this aim, our approach can be divided in two phases: for each frame, the projection histograms (Haritaoglu et al., 1998) of each person are computed and compared with the probabilistic projection maps stored for each posture during the training phase; then, the obtained posture is further validated exploiting the information extracted by a tracking module in order to take into account the reliability of the classification of the first phase. Moreover, the tracking algorithm is used to handle occlusions, making the system particularly robust even in indoors environments. Extensive experimental results demonstrate a promising average accuracy of more than 95% in correctly classifying human postures, even in the case of challenging conditions.

People reidentification in surveillance and forensics: A survey

The field of surveillance and forensics research is currently shifting focus and is now showing an ever increasing interest in the task of people reidentification. This is the task of assigning the same identifier to all instances of a particular individual captured in a series of images or videos, even after the occurrence of significant gaps over time or space. People reidentification can be a useful tool for people analysis in security as a data association method for long-term tracking in surveillance. However, current identification techniques being utilized present many difficulties and shortcomings. For instance, they rely solely on the exploitation of visual cues such as color, texture, and the object’s shape. Despite the many advances in this field, reidentification is still an open problem. This survey aims to tackle all the issues and challenging aspects of people reidentification while simultaneously describing the previously proposed solutions for the encountered problems. This begins with the first attempts of holistic descriptors and progresses to the more recently adopted 2D and 3D model-based approaches. The survey also includes an exhaustive treatise of all the aspects of people reidentification, including available datasets, evaluation metrics, and benchmarking.

3DPeS: 3D people dataset for surveillance and forensics

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.

Probabilistic people tracking for occlusion handling

This work presents a novel people tracking approach, able to cope with frequent shape changes and large occlusions. In particular, the tracks are described by means of probabilistic masks and appearance models. Occlusions due to other tracks or due to background objects and false occlusions are discriminated. The tracking system is general enough to be applied with any motion segmentation module, it can track people interacting each other and it maintains the pixel assignment to track even with large occlusions. At the same time, the update model is very reactive, so as to cope with sudden body motion and silhouettes shape changes. Due to its robustness, it has been used in many experiments of people behavior control in indoor situations.

Video Surveillance Online Repository (ViSOR): an integrated framework

The availability of new techniques and tools for Video Surveillance and the capability of storing huge amounts of visual data acquired by hundreds of cameras every day call for a convergence between pattern recognition, computer vision and multimedia paradigms. A clear need for this convergence is shown by new research projects which attempt to exploit both ontology-based retrieval and video analysis techniques also in the field of surveillance. This paper presents the ViSOR (Video Surveillance Online Repository) framework, designed with the aim of establishing an open platform for collecting, annotating, retrieving, and sharing surveillance videos, as well as evaluating the performance of automatic surveillance systems. Annotations are based on a reference ontology which has been defined integrating hundreds of concepts, some of them coming from the LSCOM and MediaMill ontologies. A new annotation classification schema is also provided, which is aimed at identifying the spatial, temporal and domain detail level used. The ViSOR web interface allows video browsing, querying by annotated concepts or by keywords, compressed video previewing, media downloading and uploading. Finally, ViSOR includes a performance evaluation desk which can be used to compare different annotations.

An integrated multi-modal sensor network for video surveillance

To enhance video surveillance systems, multi-modal sensor integration can be a successful strategy. In this work, a computer vision system able to detect and track people from multiple cameras is integrated with a wireless sensor network mounting PIR (Passive InfraRed) sensors. The two subsystems are briefly described and possible cases in which computer vision algorithms are likely to fail are discussed. Then, simple but reliable outputs from the PIR sensor nodes are exploited to improve the accuracy of the vision system. In particular, two case studies are reported: the first uses the presence detection of PIR sensors to disambiguate between an opened door and a moving person, while the second handles motion direction changes during occlusions. Preliminary results are reported and demonstrate the usefulness of the integration of the two subsystems.

A Hough transform-based method for radial lens distortion correction

The paper presents an approach for a robust (semi-)automatic correction of radial lens distortion in images and videos. This method, based on the Hough transform, has the characteristics to be applicable also on videos from unknown cameras that, consequently, can not be a priori calibrated. We approximated the lens distortion by considering only the lower-order term of the radial distortion. Thus, the method relies on the assumption that pure radial distortion transforms straight lines into curves. The computation of the best value of the distortion parameter is performed in a multi-resolution way. The method precision depends on the scale of the multi-resolution and on the Hough spaces resolution. Experiments are provided for both outdoor, uncalibrated camera and an indoor, calibrated one. The stability of the value found in different frames of the same video demonstrates the reliability of the proposed method.

HMM based action recognition with projection histogram features

Hidden Markov Models (HMM) have been widely used for action recognition, since they allow to easily model the temporal evolution of a single or a set of numeric features extracted from the data. The selection of the feature set and the related emission probability function are the key issues to be defined. In particular, if the training set is not sufficiently large, a manual or automatic feature selection and reduction is mandatory. In this paper we propose to model the emission probability function as a Mixture of Gaussian and the feature set is obtained from the projection histograms of the foreground mask. The projection histograms contain the number of moving pixel for each row and for each column of the frame and they provide sufficient information to infer the instantaneous posture of the person. Then, the HMM framework recovers the temporal evolution of the postures recognizing in such a manner the global action. The proposed method have been successfully tested on the UT-Tower and on the Weizmann Datasets.

SARC3D: a new 3D body model for people tracking and re-identification

We propose a newsimplified 3Dbody model (called SARC3D) for surveillance application, which can be created, updated and compared in real-time. People are detected and tracked in each calibrated camera, with their silhouette, appearance, position and orientation extracted and used to place, scale and orientate a 3D body model. For each vertex of the model a signature (color features, reliability and saliency) is computed from 2D appearance images and exploited for matching. This approach achieves robustness against partial occlusions, pose and viewpoint changes. The complete proposal and a full experimental evaluation are presented, using a new benchmark suite and the PETS2009 dataset.

Dynamic Pictorially Enriched Ontologies for Digital Video Libraries

This article presents a framework for automatic semantic annotation of video streams with an ontology that includes concepts expressed using linguistic terms and visual data.