城市混合交通场景中的多目标参数检测

Multiple Objects Parameter Detection in Urban Mixed Traffic Scene

针对城市混合交通的复杂场景图像中多目标及其参数的检测问题,提出了一种由改进的帧间差分与边缘提取相结合的算法。利用帧间差分法检测车辆的存在,对帧差图像运用统计滤波算法提取多运动目标,通过形态学方法提取并细化目标边缘,根据主边缘(轮廓)信息完成车辆定位,最终结合摄像机标定结果计算出多目标交通参数。算法避免了复杂场景的背景建模,减少了运算量。实验结果表明,该算法不仅能较为准确地检测多运动目标的参数,而且具有较强的实时性。

For the multiple objects parameter detection in urban mixed traffic scene, a new algorithm of multiple objects parameter detection based on difference-image and edge detection was developed. In the algorithm, the differenceimage was used to detect the vehicle. Meanwhile, the statistical filtering algorithm was used in order to extract the multiple moving objects in the difference-image, while the morphology method was used to extract and refine the edge. Finally, the multiple objects parameter was calculated. The algorithm avoids the modeling process of background, and reduces the calculation. The experiment result shows that the algorithm can detect the multiple objects parameter effectively, and that the algorithm has strong timeliness.