Dynamic Traffic Light Optimization and Control System
The current traffic control system is incapable since traffic signal system timers have
fixed time period to switch traffic between different phases whereas traffic density pattern
varies with time throughout the day. Due to which, some vehicles might have to wait for
long period of time. Such situation creates heavy traffic at one side of road and other side
with no traffic.To overcome this problem we have proposed a model that estimates
number of vehicles, specific phase traffic area, width of road. From the above generated
values we have optimized traffic signal cycle time and individual traffic signal phase time.
We have used Image processing ( deep learning ) and Machine Learning. For image
processing Viola Jones Haar Wavelet technique is used. It takes different form to
construct mini classifiers to detect object differentiated by neagative and positive
energies. Haar wavelet are pixel distribution that helps to classify edges in an image.
Haar cascade is been used for these wavelets to overlay on the given object and create
XML file to identify the vehicles and length of traffic. After that Machine learning is used
to predict cycle time and allocate time to the phases of traffic signal.