In the framework of developing a control system for a parking-lot, it is important to know the number of vehicles parking in the different areas of the lot.
Abstract
In the framework of developing a control system for a parking-lot, it is important to know the number of vehicles parking in the different areas of the lot. In addition, it is important to identify “sneaking” vehicles: those entering the parking lot when the barrier is already open, without having passed their magnetic identification card.
The problem
In order to achieve this goal, we are required to be able to:
- Count vehicles moving in a certain lane, including their direction
- Identify status of the barrier – open or closed
All the above must work in REALTIME.
The solution
In this project we developed and implemented a system that fulfills these requirements, using video taken from a high point above the entrance to the lot.
Our system identifies entry and exit of vehicles to/from the lot, and is capable of locating the car’s lane, thus counting the number of cars in the different areas of the parking lot. We developed algorithms for identification of vehicles, counting vehicles during motion, and identifying the state of the barrier in the entrance to the lot. Identifying barrier state can be done using the Radon transform, which finds straight lines in an image. The closed barrier is a distinct line in the image. Another system we implemented is using a difference image of a small area around the edge of the barrier:
Counting cars is done using the labeling method – in the binairzed difference image (comparing to a reference image), adjacent pixels are grouped together as objects. We define 2 “points of interest” in every lane, and look at the size of the objects “stepping” on these points.
If the size exceeds a predefined threshold, the object is identified as a vehicle.
Finding the direction of movement of the cars is done by looking at the order in which the cars “stepped” on the different points. The counting method itself was implemented using buffers that can a count large number of cars.
Tools
The project was developed for the parking lot of the Faculty of Electrical Engineering at the Technion-IIT, and was implemented in MatLab. The system was built from a generic perspective, so that it can be adapted relatively easily to other parking lots.
Conclusions
The developed system gave very good results in simulation. Counting cars, detecting barrier state and identifying sneaking cars is done successfully. See demo video.
Acknowledgment
We would like to thank our mentor Johanan Erez, the lab’s chief engineer, for his help and support during all stages of the project.
We are also grateful to the Ollendorf Minerva Center Fund for supporting this project.