Image processing techniques are now considered flexible and practical for collecting and analyzing road traffic data. However, traditional image processing techniques based on grey scale images have not provided good results. In this paper we introduce a new technique, which is based on colour motion segmentation and split-merge segmentation approaches. First we use motion segmentation to determine the rough position of moving vehicles in a sequence of images. Then we apply the split-merge segmentation on the colour images. In this way we need not process the whole image, which saves computation time. Instead of determining the threshold value manually, which is the case in most vision-based traffic systems, we use an adaptive threshold to automatically choose the threshold value for split-merge method. We also classify the vehicle into 4 categories based on the feature of contour. The experiment results show that this method is quite promising.
[1]
Patrick C. Chen,et al.
Segmentation by texture using a co-occurrence matrix and a split-and-merge algorithm☆
,
1979
.
[2]
Yean‐Jye Lu,et al.
Vehicle Classification Using Infrared Image Analysis
,
1992
.
[3]
Philippe Bolon,et al.
A region-region and region-edge cooperative approach of image segmentation
,
1994,
Proceedings of 1st International Conference on Image Processing.
[4]
W Pan,et al.
Automatic vehicle classification system
,
1991
.
[5]
William C. Schwartz.
Laser vehicle detector/classifier
,
1995,
Other Conferences.
[6]
Anil K. Jain,et al.
Contour extraction of moving objects in complex outdoor scenes
,
1995,
International Journal of Computer Vision.