Research on Vehicle Traffic Counting Using Area-Based Method

In this paper, we focus on studying the area-based method for vehicle traffic counting. During the algorithm development process, we utilized fundamental image processing techniques related to vehicle image detection, including grayscale conversion (implemented through weighted average of RGB channels), filtering algorithms (such as Gaussian and median filters for noise reduction), image enhancement (using histogram equalization), and sharpening techniques (employing Laplacian or Sobel operators). We conducted comparative studies on the algorithmic implementations of these techniques. Additionally, we applied image segmentation technologies including threshold segmentation (using Otsu's method for automatic threshold determination), edge detection algorithms (Canny and Sobel operators), morphological operations (dilation and erosion for shape refinement), along with vehicle detection and extraction algorithms. Finally, we designed a real-time and reliable counting algorithm that incorporates frame differencing and blob analysis to meet our research objectives.