Huge emissions of smoke from vehicles elevate the risk of respiratory infections, life-threatening diseases like cancer, heart problems, pulmonary diseases, etc. It is an extremely serious task to detect the smoky veh...
详细信息
Huge emissions of smoke from vehicles elevate the risk of respiratory infections, life-threatening diseases like cancer, heart problems, pulmonary diseases, etc. It is an extremely serious task to detect the smoky vehicles for ensuring proper maintenance and limiting hazardous emissions. Smoky vehicle emissions continue to be a critical challenge and an alarming threat in densely populated areas where the quality of air is poor. This paper proposes an automated method for detecting smoky vehicles from traffic surveillance videos using the structural feature extraction technique. To detect the vehicles, the universal background subtraction algorithm Vibe is used, and based on some rules, non-vehicle objects are removed. The features are extracted from the smoky and non-smoky vehicles using a modified structural co-occurrence matrix. Finally, the random forest classifier is used to classify non-smoky and smoky vehicles. The results evince that the proposed method achieves an overall accuracy of 96.50% and outperforms the other state-of-the-art feature extraction methods.
暂无评论