Confidence-Based Traffic Density Estimation Using Square Binary Patterns
International journal of intelligent engineering and systems · 2024
Traffic density estimation is very important in the field of Intelligent Transportation Systems thanks to the growing number of road users.This research proposes a new method for estimating traffic density using road surveillance cameras.The approach involves two main steps: block occupancy prediction and traffic density classification.The aggregation of the block occupancy prediction which results in the so-called density coefficient is the first contribution of this paper.The second contribution involves the use of semi-supervised mechanism for classifying traffic density states.Experimental results show that the proposed approach successfully obtained high accuracy for classifying block occupancies on both the proposed dataset and the existing public datasets with the best score of 99.8%.The traffic state classification result itself is satisfactory as it achieved the classification rate of up to 94%.Furthermore, the use of SBP features allows the system to work up to 3.8 times faster than LBP.