Research On The E-V Combined Indoor Localization Algorithms

The localization technology attracts more and more attention with thepopulation of3G communication technology, smart terminal as well aslocation based service (LBS). Real time localization has become thefundamental technique of multiple high level applications such as traffic,commercial services, logistics and personality services. The world famousGPS service provides one solution for outdoor localization, which has beenwidely used in many fields. In indoor environment, however, GPS cannotreach high accuracy as it performs outdoor with the influence of fading andobstruction, which is quite important in buildings to determine the differencebetween floors and roomsWLAN indoor localization is one of the most used indoor localizationscheme. WLAN system as well as its hardware has been deployed in mostpublic indoor space, which means no additional device is needed to realizethe localization system with WiFi.Moreover, WiFi indoor localization withfingerprint algorithm has excellent effect on solving the complex wirelesselectromagnetic wave transportation in complex indoor environment, whichavoid the defect of localization scheme with triangulation. This thesisdesigns and realizes a WiFi fingerprint localization system, which uses multi-direction fingerprint database and smooth mechanism to improve accuracy onthe base of traditional fingerprint localization system. This thesis alsopropose an differential fingerprint database to meet the localization need fordifferent localization terminal.On the other hand, localization with the help of visual tracking is in rapid growth. Visual localization system with the help of people detect,coordinate calibration and visual tracking decrease the localization errorgreatly in contrast to electrical localization. Nevertheless, one defect limits itsapplication: visual localization cannot determine the identity of target inimage. To solve this problem as well as improves the accuracy of electricallocalization, this thesis proposes an algorithm to combine electroniclocalization and visual localization. We uses KM algorithm to make perfectmatch between E-loc result and V-Loc result. This method sacrifices part ofV-Loc accuracy to increase E-Loc accuracy as well as clarify the identity ofV-Loc target. The analysis of this system shows that precise accuracy stillkeeps with the appearance of interference target.