The increasing level of traffic congestion poses a huge challenge to the transportation sector and causes environmental and health hazards. To address this challenge, a better understanding of the traffic system is required. Towards this effort, it is important to collect and analyze real-time traffic data. Accurate vehicle detection and classification in real-time is a challenging problem for engineers and researchers working in the field of transportation engineering, especially under heterogeneous and weak lane-disciplined traffic conditions. To detect vehicle presence in lane-based traffic, one of the widely used traffic detectors is the inductive loop detector (ILD). For weak lane-disciplined and heterogeneous traffic, researchers have developed a modified solution, named as multiple inductive loop detector (MILD) system. In order to extract classified vehicle count from MILD data, it is required to detect vehicle presence and segment the signature of different vehicles. This study proposes a multivariate data analysis framework for the detection and segmentation of vehicle signatures from the acquired data, without significant manual intervention. The major challenge in this process is the coupling of the multi-dimensional loop data, due to cross-talk across the loops. To address this, principal component analysis (PCA) is used with the additional benefit of dimensionality reduction. The results show that the developed algorithm achieved an average vehicle count accuracy of 90.38 %. Next, a K-nearest neighbor (KNN) based vehicle classifier is proposed to extract the classified vehicle count. Statistical measures such as maximum amplitude, length of vehicle signature, mean, and variance are used as features to perform the classification operation. Overall, a classification accuracy of 92 % is achieved with an F1 score of 0.8, 0.5, and 0.96 for vehicle classes Bus, Car, and Two-wheelers respectively. A more detailed study is required in order to select suitable features, to achieve good accuracy for all classes of vehicles.

Keywords: Traffic monitoring, inductive loop detectors (ILD), vehicle detection, segmentation, classification.