Study on the Technology of Vehicle Flow Statistics Based on Video Images
Chang Liu, Xiaogang Zhang, Baiyu Ma, Jun Wei Mao · 2025
Vehicle traffic flow statistics is a critical research topic in the field of intelligent transportation systems (ITS). Traditional traffic flow statistics methods suffer from issues such as long detection times and insufficient accuracy. This study leverages the OpenCV platform to implement video image-based vehicle detection, tracking, and statistical counting. Vehicle detection is achieved using the background subtraction method, while vehicle tracking combines centroid features with a Kalman filter. Traffic flow counting is realized through the design of counting lines. Experimental results demonstrate that for video images with a resolution of$352 \times 240$and a frame rate of 25 frames per second (fps), the average accuracy rates for both vehicle detection and counting exceed$\mathbf{9 5 \%}$, meeting basic real-time requirements.