Haupt-Navigation

Page Content

Marie-Neige Garcia

: © Copyright??

Research Field:

- Quality and Usability

Research Topics:

- Video and Audio-Visual Quality Assessment
- Quality modeling

Biography

Since 2006: PhD Student at Technische Universität Berlin / Deutsche Telekom Laboratories

2004-2006: Project manager at ELDA (Evaluation and Language Resources Distribution Agency) , Speech evaluation department - Paris, France

Project manager in the EVALDA/EvaSy

campaign (Evaluation of speech synthesis systems in French)
In charge of the evaluation of

Text-to-speech systems in the European project TC-STAR (speech-to-speech

translation)
In charge of the evaluation of video

technologies in the project CHIL (Pervasive Interaction Computing)

2003: Master’s Degree in Engineering at ISEP (Electronics Superior Institute of Paris); areas of study: informatics, telecommunications, electronics

Address

Quality and Usability Lab
Deutsche Telekom Laboratories
TU Berlin
Ernst-Reuter-Platz 7
D-10587 Berlin, Germany

: © Copyright??

Publications

No-Reference bit stream model for video quality assessment of H.264/AVC video based on packet loss visibility
Citation key	argyropoulos2011b
Author	Argyropoulos, Savvas and Raake, Alexander and Garcia, Marie-Neige and List, Peter
Title of Book	IEEE Int. Conf. on Acoustics, Speech, and Signal Processing (ICASSP)
Pages	1169–1172
Year	2011
ISBN	978-1-4577-0537-3
ISSN	1520-6149
Location	Prague, Czech Republic
Address	Prague, Czech Republic
Month	may
Note	electronic/online
Publisher	IEEE
How Published	poster
Abstract	In this paper, a no reference bit stream model for quality assessment of SD and HD H.264/AVC video sequences based on packet loss visibility is proposed. The method considers the impact of network impairments on human perception and uses the visibility of packet losses to predict objective scores. Also, a new subjective experiment has been designed to provide insight into the perceptual effect of degradations caused by transmission errors. The proposed algorithm extracts a set of features from the received bit stream. Then, the visibility of each packet loss event is determined by classifying the extracted features using a Support Vector Machines classifier. Finally, analytical expressions are developed to account for visual degradation due to compression and channel induced distortion based on the outcome of the visibility classifier. The evaluation demonstrates the validity of the proposed method.