Vision-Based Counting of Moving Vehicles Using Catcher Algorithm

Owing to the increasing social concerns caused by high traffic density in developing countries, governments should incorporate intelligent traffic systems to regulate traffic.Vehicle detection, tracking, and counting play important roles in vision-based intelligent traffic systems.Robustness to errors (i.e., missed and doubled counts) is essential in traffic systems.Although the existing Marker algorithm can prevent missed counts, it provides a doubled count when two or more binary objects come from a vehicle.This study applies and analyzes the Catcher algorithm to three highway lanes to prove whether it can prevent the double count occurrence.The results showed that compared to the Marker algorithm, the Catcher algorithm demonstrates better accuracy for the tested cases.Its initial accuracy reached 94.26% and increased further after adjusting the size of the region of interest.In conclusion, the Catcher algorithm can prevent a double count by rounding down when two or more binary objects originate from a vehicle.