High-Accuracy Vehicle Detection in Different Traffic Densities Using Improved Gaussian Mixture Model with Cuckoo Search Optimization

Background subtraction plays a critical role in computer vision, particularly in vehicle detection and tracking. Traditional Gaussian Mixture Models (GMM) face limitations in dynamic traffic scenarios, leading to inaccuracies. This study proposes an Improved GMM with adaptive time-varying learning rates, exponential decay, and outlier processing to enhance performance across light, moderate, and heavy traffic densities. The model's parameters are automatically optimized using the Cuckoo Search algorithm, improving adaptability to varying environmental conditions. Validated on the ChangeDetection.net 2014 dataset, the Improved GMM achieves superior precision, recall, and F-measure compared to existing methods. Its consistent performance across diverse traffic scenarios highlights its effectiveness for real-time traffic flow analysis and vehicle detection applications.

@article{Aris2025,
  title     = {High-Accuracy Vehicle Detection in Different Traffic Densities Using Improved Gaussian Mixture Model with Cuckoo Search Optimization},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {16},
  number    = {1},
  year      = {2025},
  publisher = {The Science and Information Organization},
  author    = {Nor Afiqah Mohd Aris and Siti Suhana Jamaian},
  doi       = {10.14569/IJACSA.2025.01601100},
  url       = {https://doi.org/10.14569/IJACSA.2025.01601100}
}