基于强度峰值的运动车辆识别方法

针对雾霾天气下照度不够、视频图像背景不断变化导致雾霾天图像质量下降、前景目标不易检测识别等现象进行了分析,解决了传统光流法无法有效单独检测跟踪目标的问题。在分析研究传统车辆识别思路及方法的基础上,提出一种光流估计与强度峰值特征提取相结合的检测方法,可以适应跟踪过程中目标特征和背景的不断变化,有效解决雾霾天气条件下车辆检测的鲁棒性问题。最后,通过实验验证了方法的有效性。

A Method of Moving Vehicles Recognition Through Intensity Peak

In fog-haze weather, car video image sequence capture suffers serious degradation owing to the low visibili- ty and the test results are often not ideal. Based on the analysis of the traditional methods of vehicle identification, this paper discussed the issues leading to the difficulties in detecting foreground object and the degradation of image quality under fog-haze weather,namely, the insufficient degree of illumination, the constantly changing image back- ground. A method of detecting moving vehicles was proposed through the combination of optical flow estimation and peak features extraction. This method is most applicable to conditions of the everrehanging process of tracking target and background and improves robustness of intelligent vehicle detection under fog haze weather conditions. The experimental results have proved the effectiveness of this method.