Conference paper Open Access

# Live Demostration: Low Power Vision Sensor with Robust Dynamic Background Rejection

Zou, Yu; Gottardi, Massimo; Perenzoni, Matteo

### DataCite XML Export

<?xml version='1.0' encoding='utf-8'?>
<identifier identifierType="URL">https://zenodo.org/record/3238271</identifier>
<creators>
<creator>
<creatorName>Zou, Yu</creatorName>
<givenName>Yu</givenName>
<familyName>Zou</familyName>
<affiliation>Fondazione Bruno Kessler</affiliation>
</creator>
<creator>
<creatorName>Gottardi, Massimo</creatorName>
<givenName>Massimo</givenName>
<familyName>Gottardi</familyName>
<affiliation>Fondazione Bruno Kessler</affiliation>
</creator>
<creator>
<creatorName>Perenzoni, Matteo</creatorName>
<givenName>Matteo</givenName>
<familyName>Perenzoni</familyName>
<affiliation>Fondazione Bruno Kessler</affiliation>
</creator>
</creators>
<titles>
<title>Live Demostration: Low Power Vision Sensor with Robust Dynamic Background Rejection</title>
</titles>
<publisher>Zenodo</publisher>
<publicationYear>2018</publicationYear>
<dates>
<date dateType="Issued">2018-05-27</date>
</dates>
<resourceType resourceTypeGeneral="Text">Conference paper</resourceType>
<alternateIdentifiers>
<alternateIdentifier alternateIdentifierType="url">https://zenodo.org/record/3238271</alternateIdentifier>
</alternateIdentifiers>
<relatedIdentifiers>
<relatedIdentifier relatedIdentifierType="DOI" relationType="IsIdenticalTo">10.1109/ISCAS.2018.8351881</relatedIdentifier>
</relatedIdentifiers>
<rightsList>
<rights rightsURI="info:eu-repo/semantics/openAccess">Open Access</rights>
</rightsList>
<descriptions>
<description descriptionType="Abstract">&lt;p&gt;Commercial cameras are mainly targeted to visual tasks, where resolutions and image quality are the most important parameters. However, in several applications, such as surveillance and monitoring, they are not very energy efficient. In fact, they continuously acquire images, forcing an external processor to process images in real-time, looking for events to occur in the scene that may rarely or even never happen. This causes the system to process a large amount of data uselessly, burning high power. Embedding low-level image processing on-chip would make the system more efficient, dispatching only salient features, thus reducing data bandwidth and the off- chip burden of computation as well. During this live demonstration, we will present an always-on QVGA (320 x 240 pixels) visual sensor embedding unusual motion detection [1] targeted to surveillance applications. The sensor detects anomalous motion in the scene and dispatches the grey-scale image and related bitmap of event at 15 fps with 1.6mW power consumption. Differently from other sensors relying on frame- difference to detect motion, the presented sensor embeds a two-thresholds dynamic background subtraction algorithm [2], which allows noisy background (such as swaying vegetation and rippling waves) to be suppressed in a more robust way, thus making the sensor to be suitable also for outdoor applications.&lt;/p&gt;</description>
</descriptions>
<fundingReferences>
<fundingReference>
<funderName>European Commission</funderName>
<funderIdentifier funderIdentifierType="Crossref Funder ID">10.13039/501100000780</funderIdentifier>
<awardNumber awardURI="info:eu-repo/grantAgreement/EC/H2020/653355/">653355</awardNumber>
<awardTitle>FOREnsic evidence gathering autonomous seNSOR</awardTitle>
</fundingReference>
</fundingReferences>
</resource>

23
33
views