ABSTRACT The Parliament of Andalusia records all the parliamentary sessions as well as generates files with the exact transcription of the files. With these two types of media, a search engine, starting from a user’s query, would return the relevant documents for that query, but also a link to the corresponding portion of the video where the speech is played. To achieve this goal, a previous tool to segment the videos and later to annotate or synchronize text and video must be developed. In this paper we describe the tool for segmentation and annotation, more specially the simple but effective and efficient segmentation algorithm developed exclusively for the special features of the videos from the Parliament of Andalusia. KEYWORDS Video segmentation, annotation, parliamentary sessions. 1. INTRODUCTION AND MOTIVATION One of the main objectives of a democracy is that citizens know what their representatives in the parliament are dealing with in each moment. In this line, national and regional parliaments have to spread the works developed in these chambers of members of parliament in order to make public all the matters being discussed. So the Parliament of Andalusia, the southern region of Spain, generates a group of electronic documents in PDF format called session diaries, published in the www.parlamentodeandalucia.es site. They store all the discussed matters in every session and the participations of the members of parliament for these matters. These session diaries belong to different legislatures, composed of, at most, four years of politics activity. Nowadays, since the creation of the Parliament of Andalusia in 1982, there have been seven legislatures with more than 1500 PDF documents. Moreover, the sessions are recorded in video, so additionally to the transcriptions, the digital library of the Parliament is complemented with the videos. In the session diaries, and therefore, in the videos, we can find all the participations of the members of parliament, and also all the agreements achieved in the plenary sessions of the Permanent and Commission Delegation passing laws or celebrating informative sessions with members of the regional Government. The session diary and its corresponding video are published in the website after the meetings of the deputies. These documents could be accessed through by means of a search engine that works with a representation in XML of the PDF documents where the internal structure of the session diaries could be exploited. Then the user formulates a query and gets the relevant documents (Baeza-Yates, Ribeiro-Neto, 2001), or parts of them (Chiaramella, 2001). But, this is not the case of the videos, which may be accessed by date, basically. There is no link between the document of the session diary and the video. But it could be very useful for the user that when she/he retrieves a relevant document (the text), or a portion of it, she/he could watch the associated video at the same time. Then the structured information retrieval field (Chiaramella, 2001; de Campos et al., 2006) gives the possibility of making the decision of determining the type of XML element containing relevant information, so the user does not have to inspect the whole document to find the requested information. But also, the user could watch only the portion of the video in real time. This feature could be an added value for the search engine.
[1]
Stefanos D. Kollias,et al.
A Stochastic Framework for Optimal Key Frame Extraction from MPEG Video Databases
,
1999,
Comput. Vis. Image Underst..
[2]
Michel Barlaud,et al.
Video segmentation using active contours on a group of pictures
,
2002,
Proceedings. International Conference on Image Processing.
[3]
Andrew W. Fitzgibbon,et al.
Automatic Video Segmentation using Spatiotemporal T-Junctions
,
2006,
BMVC.
[4]
Amit Jain,et al.
A Fast Method for Textual Annotation of Compressed Video
,
2002,
ICVGIP.
[5]
Afzal A. Godil.
VISA: Video Segmentation and Annotation
,
2004
.
[6]
Guoliang Fan,et al.
Combined key-frame extraction and object-based video segmentation
,
2005,
IEEE Transactions on Circuits and Systems for Video Technology.
[7]
Janko Calic,et al.
Spatial analysis in key-frame extraction using video segmentation
,
2004
.
[8]
Noel E. O'Connor,et al.
Temporal video segmentation for real-time key frame extraction
,
2002,
2002 IEEE International Conference on Acoustics, Speech, and Signal Processing.
[9]
Janko Calic,et al.
Efficient key-frame extraction and video analysis
,
2002,
Proceedings. International Conference on Information Technology: Coding and Computing.
[10]
Luis M. de Campos,et al.
Garnata: An information retrieval system for structured documents based on probabilistic graphical models
,
2006
.
[11]
José M. Palomares,et al.
New edge-based feature extraction algorithm for video segmentation
,
2003,
IS&T/SPIE Electronic Imaging.
[12]
Yu-Jin Zhang,et al.
Video Segmentation and Key Frame Extraction with
,
2008
.
[13]
Azriel Rosenfeld,et al.
Compressed Domain Video Segmentation
,
1996
.
[14]
Feng Wu,et al.
Automatic video segmentation using a novel background model
,
2002,
2002 IEEE International Symposium on Circuits and Systems. Proceedings (Cat. No.02CH37353).
[15]
Yves Chiaramella,et al.
Information Retrieval and Structured Documents
,
2000,
ESSIR.
[16]
Christos Faloutsos,et al.
Compressed-domain video indexing techniques using DCT and motion vector information in MPEG video
,
1997,
Electronic Imaging.