188宝金博页面版

Wavelet Transform Based Vehicle Detection from Sensorsfor Bridge Weigh-in-MotionJuan Alegre-SanahujaNational Institute ofInformaticsTokyo 101-8430, Japanjuaalsa@doctor.upv.comWei LuNational Institute ofInformaticsTokyo 101-8430, Japanrogi@nii.ac.jpAtsuhiro TakasuNational Institute ofInformaticsTokyo 101-8430, Japantakasu@nii.ac.jpABSTRACTA variety of technologies can be applied to the collectionand analysis of traf f i c data in smart cities. Vehicle detec-tion, which is a fundamental aspect of traf f i c analysis, can beachieved by various technologies such as surveillance videosand loop detectors. This paper proposes a vehicle detectionmethod that uses a set of sensors for bridge weigh-in-motion,which is an in situ nonintrusive method that avoids the dis-advantages of other systems. In a practical implementationof this method, where data streams from sensors have to beprocessed in real time, we found vehicle-detection inaccura-cies caused by the characteristics of signals from the sensors.To address these problems, we propose a simple and ef f i cientmethod that uses two sensors and a wavelet transform. Ourmethod improves the system accuracy by comparing the re-sults of a robust wavelet-transform peak-detection techniqueapplied to the signal streams from the two sensors. Exper-imental results demonstrate the high performance of thismethod, which can meet the accuracy requirements of real-time scenarios.CCS Concepts•Applied computing → Engineering; •Hardware →Sensor applications and deployments; Digital signalprocessing; •Information systems → Data streaming;KeywordsUrban computing; Traf f i c data analysis; Real time; Wavelettransform; BWIM.1. INTRODUCTIONVarious technologies can be applied to the computationsused in smart cities to collect traf f i c data. These data mayPermission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copiesare not made or distributed for prof i t or commercial advantage and thatcopies bear this notice and the full citation on the f i rst page. Copy-rights for components of this work owned by others than ACM mustbe honored. Abstracting with credit is permitted. To copy otherwise,or republish, to post on servers or to redistribute to lists, requires priorspecif i c permission and/or a fee. Request permissions from Permis-sions@acm.org.SAC 2016,April 04-08, 2016, Pisa, Italyc ? 2016 ACM. ISBN 978-1-4503-3739-7/16/04...$15.00DOI: http://dx.doi.org/10.1145/2851613.2851656include traf f i c counts, vehicle classif i cations, vehicle veloci-ties and travel times, with the aim of improving traf f i c in-formation, enhancing public safety, informing about traf f i cstates and reducing congestion. There are resultant benef i tsand savings for citizens, motor carriers, highway operatorsand government agencies [1].These technologies can be split into “in situ” approaches,which use detectors located along the roadside [2], and probe-car systems [9], which use data collected by traveling vehi-cles f i tted with sensors. The “in situ” technologies can beclassif i ed further into intrusive or in-roadway, where embed-ded or taped detectors and sensors are installed on or belowthe road surface, and nonintrusive or over-roadway, wherethe sensors are mounted above or alongside the roadway [2].Among the “in situ” technologies, we focus on the bridgeweigh-in-motion (BWIM) system for the source of vehicledetection because it is nonintrusive, undetectable by passingtraf f i c, avoids any road closure for installation and mainte-nance, can be installed without interference with the roadsurface, is easily portable and can work while vehicles aretraveling at normal highway speed [7]. Furthermore, whencounting the passing vehicles, this system avoids the privacyissues of video systems [14].In a practical implementation of a BWIM system, variousproblems can appear, related to noisy signals from sensors,which leads to undetected vehicles and false-positives (nonex-istent vehicles), or related to the static component of thesignals, which can produce erroneous velocity measurements[8].In this work, we present a new method based on the wavelettransform that uses the data streams from two sensors. Themethod processes in real time the streams of data from twosensors installed at the bridge, which solves the vehicle-detection and velocity-estimation problems caused by thecharacteristics of the sensor signals. It also enables real-timeprocessing of the data streams using a fast algorithm for real-time applications. The main contributions of this paper arethat it introduces the wavelet-transform technique for ro-bust vehicle detection against noise and utilizes two sensorsto improve detection accuracy.For the implementation and testing of our proposed system,we utilized the signals from sensors installed at a bridge ona highway located in the suburbs of Tokyo.935