Abstract
Currently, sensors embedded in automobiles for surveillance purpose are mainly composed of video based and/or acceleration based sensors. However, there are surveillance situations wherein image and acceleration information obtainable from vehicles may not be sufficient for recognizing surrounding abnormal situations. Such applicable events considered in this investigation include screaming, police car sirens, and noise of glass breaking observable in both driving and parked scenarios. In this paper, a two-step event recognition system is proposed. The first step is event detection through a threshold in high-frequency acoustic band, based on an assumption that the events considered here have greater acoustic energy concentrated in the high-frequency region compared to the background noise coming from the vehicle engine. The next step is to precisely classify the detected event robust from surrounding false positives. We present a set of features which are beat spectrum, spectral flux and their combinations. To enhance the classification performance, we employed confidence measures derived from time normalized log-likelihood score.
Original language | English |
---|---|
Title of host publication | 5th Biennial Workshop on Digital Signal Processing for In-Vehicle Systems, DSP 2011 |
Publisher | European Association for Signal Processing (EURASIP) |
Publication status | Published - 2011 Jan 1 |
Event | 5th Biennial Workshop on Digital Signal Processing for In-Vehicle Systems, DSP 2011 - Kiel, Germany Duration: 2011 Sep 4 → 2011 Sep 7 |
Other
Other | 5th Biennial Workshop on Digital Signal Processing for In-Vehicle Systems, DSP 2011 |
---|---|
Country | Germany |
City | Kiel |
Period | 11/9/4 → 11/9/7 |
Fingerprint
Keywords
- Acoustic event detection and classification
- Confidence measure
- Feature combination
ASJC Scopus subject areas
- Signal Processing
- Automotive Engineering
Cite this
Audition based event detection and classification for vehicular surveillance. / Boo, Jongcheol; Choi, Woohyun; Rho, Jinsang; Han, David K.; Ko, Hanseok.
5th Biennial Workshop on Digital Signal Processing for In-Vehicle Systems, DSP 2011. European Association for Signal Processing (EURASIP), 2011.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Audition based event detection and classification for vehicular surveillance
AU - Boo, Jongcheol
AU - Choi, Woohyun
AU - Rho, Jinsang
AU - Han, David K.
AU - Ko, Hanseok
PY - 2011/1/1
Y1 - 2011/1/1
N2 - Currently, sensors embedded in automobiles for surveillance purpose are mainly composed of video based and/or acceleration based sensors. However, there are surveillance situations wherein image and acceleration information obtainable from vehicles may not be sufficient for recognizing surrounding abnormal situations. Such applicable events considered in this investigation include screaming, police car sirens, and noise of glass breaking observable in both driving and parked scenarios. In this paper, a two-step event recognition system is proposed. The first step is event detection through a threshold in high-frequency acoustic band, based on an assumption that the events considered here have greater acoustic energy concentrated in the high-frequency region compared to the background noise coming from the vehicle engine. The next step is to precisely classify the detected event robust from surrounding false positives. We present a set of features which are beat spectrum, spectral flux and their combinations. To enhance the classification performance, we employed confidence measures derived from time normalized log-likelihood score.
AB - Currently, sensors embedded in automobiles for surveillance purpose are mainly composed of video based and/or acceleration based sensors. However, there are surveillance situations wherein image and acceleration information obtainable from vehicles may not be sufficient for recognizing surrounding abnormal situations. Such applicable events considered in this investigation include screaming, police car sirens, and noise of glass breaking observable in both driving and parked scenarios. In this paper, a two-step event recognition system is proposed. The first step is event detection through a threshold in high-frequency acoustic band, based on an assumption that the events considered here have greater acoustic energy concentrated in the high-frequency region compared to the background noise coming from the vehicle engine. The next step is to precisely classify the detected event robust from surrounding false positives. We present a set of features which are beat spectrum, spectral flux and their combinations. To enhance the classification performance, we employed confidence measures derived from time normalized log-likelihood score.
KW - Acoustic event detection and classification
KW - Confidence measure
KW - Feature combination
UR - http://www.scopus.com/inward/record.url?scp=84899513591&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84899513591&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:84899513591
BT - 5th Biennial Workshop on Digital Signal Processing for In-Vehicle Systems, DSP 2011
PB - European Association for Signal Processing (EURASIP)
ER -