论文部分内容阅读
视频交通参数检测中的车辆分割需要准确地检测与车辆连在一起的阴影和车灯产生的亮斑。一般地,路面与车辆的图像在灰度结构上存在显著差异。本文推导了图像在光照变化情况下的一种灰度结构——极点与极性分布图,并提出了一个基于该分布图的车辆阴影与亮斑检测算法。该算法能够精确地检测车辆阴影,车灯照射产生的路面亮斑和因其他原因被误为车辆的路面。该方法使用对光照变化不敏感的灰度结构,阴影检测准确而稳定,计算量小。
Vehicle segmentation in the detection of video traffic parameters requires accurate detection of the shadows and lights produced by the vehicle in conjunction with the vehicle. In general, pavement and vehicle images in the grayscale structure there is a significant difference. In this paper, we deduce a grayscale structure of the image under the condition of illumination variation - the distribution of the pole and the polarity, and propose a vehicle shadow and bright spot detection algorithm based on the distribution map. The algorithm accurately detects vehicle shades, bright spots on the road due to headlight exposure and road surfaces that have been mistaken for vehicles for other reasons. The method uses a grayscale structure that is insensitive to changes in illumination, the shadow detection is accurate and stable, and the calculation is small.