Detection of local/non-local vehicles from AVI (Automatic Vehicle Identification) and facilitating City parking/routing plans
Publication Date: 2016-Mar-15
The IP.com Prior Art Database
Detection of local/non-local vehicles from AVI (Automatic Vehicle Identification) and facilitating City parking/routing plans Many cities are unable to do this as vehicles registered elsewhere may be freely used by the residents (in some countries this is the case) thus making it impossible to look at a license plate and say whether this is a local vehicle or a non-local vehicle.
Page 01 of 5
Detection of local
Detection of local/
While studying traffic and designing better transportation systems, cities face some interesting challenges. Especially the ones with a lot of tourist traffic. Many cities have to design their transportation systems (public transportation, roads and parking) to cater to both citizens (regular traffic) and visitors (occasional traffic). For example, the visitors may be coming in large numbers when there are games or some special events taking place in the city. For city transportation planners one of the challenges is to be able to distinguish regular citizen induced traffic from visitor-induced traffic. Many cities are unable to do this as vehicles registered elsewhere may be freely used by the residents (in some countries this is the case) thus making it impossible to look at a license plate and say whether this is a local vehicle or a non-local vehicle.
We suggest an innovative way to solve this problem.
Here is the method:
The AVI ( Automatic Vehicle Identification) cameras placed in different intersections of the city take pictures/videos of vehicles and identify the license plate numbers. Usually this data is kept for a few months and then discarded by traffic enforcement. There is usually some post processing of videos or images to recognize the license plate numbers ( especially to issue traffic violation citations). We suggest creating a historical database (table) of all vehicle license plates seen, the city zone where it is seen and the time of sighting (i.e., timestamp). This needs to be done over all AVI cameras in the city. Let's call this the SightingsTable
1. Periodically (say once a month or quarter) process the SightingsTable and generate frequency counts for each license plate during that period. If the period includes a known special event (like a sports event or a film festival in the city), it is even better.
2. Run a statistical clustering algorithm using the frequency of occurrence and vehicle license plate numbers. It produces at least two large clusters: one for vehicles with high frequency and the other for vehicles with low frequency. The high frequency cluster vehicles are the local ones (most likely) and the low frequency are non-local (non-local).
Note 1: This method is not deterministic in the sense some vehicles that are non-local may appear to be local and vice versa, but if the purpose of this scheme is to get a close approximation to local and non-local vehicles for city planning purposes, then this provides satisfactory results.
Note 2: The method can be further refined with a number of other parameters and heuristics. For example, one could also include the day of the week the vehicle has been sighted in the data collected and use it in clustering. This can give an indication of which vehicles are mostly used on weekends vs weekdays. Also by including the AVI locations near highway entries and exits, one can observe a pattern within the s...