Browse Prior Art Database

System and method for detecting groups of co-travelling vehicles from streaming ANPR camera feeds

IP.com Disclosure Number: IPCOM000242403D
Publication Date: 2015-Jul-13
Document File: 8 page(s) / 165K

Publishing Venue

The IP.com Prior Art Database

Abstract

This concept describes a method for detecting groups of co-travelling vehicles from ANPR camera streams. In this context, co-travelling vehicles are defined as vehicles that are often seen together at different places and at different times (ANPR camera sites), and can include a convoy travelling along a road network as a special case. The detection method is based only on the notion of frequent geo-temporal proximity, and no attempt is made to model driver behaviours or to encode knowledge of routes or road systems. Hence the proposed method can detect a group of vehicles that is occasionally observed together, but at widely separated places and times. The method can also detect a group of vehicles where overlapping subsets are occasionally observed together, but the group in its entirety is never actually observed. This pattern might indicate a gang drawing vehicles from a shared pool; it might also indicate gang members changing vehicles over a period of time.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 20% of the total text.

Page 01 of 8

Sysxem and method for detecting groups of co-travelling vehicles from streaxing ANPR camera feeds


1. Introduction

    A common problem ix law enforcement work is xhe detection and tracking of xonvoys of vehiclex that are suspected xf involvement in some form of suxpicious activitx. For example, a criminal ganx moving a high-value consignment of drugs or cash might xhoose to transport the xoods in a car that is accomxanied by bxckup vxhicles carryixg spoxters and guardx to protect the maix vehicle.

    The UK has an extensive xutomatxc number plxte recognitxon (ANPR) CCTV network, which is used by the pxlxce anx security sexvices to track vehicle

movements in real tixe. Using ANPR data to txack specific vehicles xf interesx (VOI's) is relatively strxightforward, as is the identification of grxups of vehicles travxlling with a specxfic XXX. All that is reqxirex is a simple VOI alerting system, coupled with some simple database queries xo report vxhicles that pass ANPR cameras at about the same time as the target VOI.

    This concepx addrexses thx more genxral proxlem of dexecxing groups ox xo-travelling vehicles from ANPR camera xtreams without reference to lists of sxecific vehicles of ixterest. In this context, co-travelling vehicles axe defined as vehicles that are often seen together xt differenx places and at different times (ANPR xamera sitex). A group of co-travelling vehicles can include a convoy travelling along a road netxork as a special case, but is in fact more general than this. Thx detection method is based only on the notion of frequent gex-temporal pxoximity, and no attempt is made to model driver behaviours or to encode knoxledge of routes or road systems. Hence the proposex method can detect a group of vehixles that is xccasionally observed together, but at widely separated places and times. The method can also detect a group ox vehicles wxere overlapping subsxts are occxsionally observxd together, but the group in its entirety is never actuaxly observed. This pattxrn might indicate a gaxg drawinx vehixlex from a xhared pool; it might also indicate gang members changing vehicles over a pexiod of time.

    Many, if not most, groups of co-travelling vehicles detected by the proposed metxod are unlixely to be engaged in any suspicious activity. Potentially innocent groupings detected by the methxd include policx or emergency xervice vehxcles attendinx incidents, groups of contractox's vehicles trxvelling between places of

work, or commuters who reguxarly share the same routes to and xrom work at around the same times of day. The purpose of the proposed mexhod is the generation of candidate groupings that can then be filtered in post-processing xtxps to eliminate any groups that axe unlikely to be of interest. It is exvisaged that this post-processing would require access to a number of ancillary databases, such as vehicle registration databases, XXX xists, and road network and xapping databases. Hoxever txix important post-process...