Marzia Corvaglia

The COST292 experimental framework for rushes summarization task in TRECVID 2008

In this paper, the method used for Rushes Summarization task by the COST 292 consortium is reported. The approach proposed this year differs significantly from the one proposed in the previous years because of the introduction of new processing steps, like repetition detection in scenes. The method starts with junk frames removal and follows with clustering and scene detection; then for each scene, repetitions are detected in order to extract once the real scene; the following step consists in face detections (faces are considered semantically relevant) and in pan, tilt and zoom detections (other camera motions are usually related to technical operations in the backstage); finally the summary is extracted.

Comparing the quality of multiple descriptions of multimedia documents

With the definition of the MPEG-7 standard, thanks to its inter-operability behaviors, it is now possible for applications to use content descriptions of a same document, coming from different sources. This implies that the overall information available at the application can be highly redundant and mechanisms for filtering the information are hence required. In this work, a general approach to descriptions integration is proposed. The idea is to manipulate information readily available from the considered descriptions to reach an accurate integration result, without having to reprocess the multimedia material. The proposed general framework has been applied at a real problem showing how segment decompositions can be integrated on the basis of dominant color descriptors series. This work introduces also the concept of reliability of descriptor extraction methods demonstrating how it can be effectively used through the proposed integration process.

CBCD based on color features and landmark MDS-assisted distance estimation

Content-Based Copy Detection (CBCD) of digital videos is an important research field that aims at the identification of modified copies of an original clip, e.g., on the Internet. In this application, the video content is uniquely identified by the content itself, by extracting some compact features that are robust to a certain set of video transformations. Given the huge amount of data present in online video databases, the computational complexity of the feature extraction and comparison is a very important issue. In this paper, a landmark based multi-dimensional scaling technique is proposed to speed up the detection procedure which is based on exhaustive search and the MPEG-7 Dominant Color Descriptor. The method is evaluated under the MPEG Video Signature Core Experiment conditions, and simulation results show impressive time savings at the cost of a slightly reduced detection performance.

Toward a multi-feature approach to Content-Based Copy Detection

Video Content-Based Copy Detection (CBCD) is an emergent research field which is targeted to the identification of modified copies of an original clip in a given dataset, e.g., on the Internet. As opposed to digital watermarking, the content itself is used to uniquely identify the video through the extraction of features that need to be robust against a certain set of predetermined video attacks. This paper advocates the use of multiple features together with detection performance estimation to construct a flexible video signature instead of a fixed, single feature based one. To combine diverse features, a normalized linear combination is also proposed. The system performance boost is evaluated through the MPEG Video Signature Core Experiment dataset and experimental results show how the proposed signature scheme can achieve impressive improvements with respect to the single feature approach.

Copy detection of immersed content in video databases

Content-Based Copy Detection techniques aim at the identification of modified copies of an original clip in a given database or on the Internet. Usually, a digital fingerprint is computed from the media itself by extracting some features which are later used for the copy detection task. In this paper we propose a system for Video Copy Detection which makes use of several low-level features. In particular we investigate the problem of the detection of copies immersed in dummy multimedia contents. By analyzing the temporal evolution of the selected features, the proposed method detects the relevant part of the query useful for the detection of the original content the query has been extracted from. Experiments conducted on the data set provided by MPEG indicate that a combined use of different features can be useful for isolating immersed copies. Moreover we can see that the various features act differently with respect to the type of modification the video query suffered.

Towards Automated Robust Vision-Based Surveillance

Summary form only given. COST is one of the longest-running instruments funded by the EU RTD Framework Program. The members of the COST 292 action (see: http://www.cost292.org/) have a long tradition of involvement in WIAMIS, dating back to the establishment of this event. Since 2005, the action has been invited to organize a special session and this tradition was continued for WIAMIS 2008. This year, the special session focused on a subset of the broad research remit of the action. Specifically, the session targets research in the field of automated visual surveillance. This focus reflects a growing interest within COST 292 participants in this increasingly important area. The goal of this special session is to help bring together researchers working on this topic, both within COST 292 and beyond, with a view to stimulating future collaborative research in this area. The widely circulated call for papers targeted submissions on a broad range of topics covered by the above mentioned focus, including visual computation, event and activity modeling and analysis, multiple stream analysis, architectures, indexing and storage, coding and applications. All papers were peer reviewed by at least three reviewers drawn from a combination of the WIAMIS Technical Program Committee and a special Program Subcommittee set up specifically for this special session. Acceptance criteria considered overall quality as well as relevance to the topics of the special session. In the end, just 5 papers were accepted for oral presentation. The accepted papers address scene modeling, event detection, object recognition (human and non-human) and fusion of multiple complementary data sources.