基于灰度渐变特征的运动车辆阴影检测

Shadow Detection of Moving Vehicle Based on Gray Gradual Characteristics

针对当前基于颜色特征的阴影检测算法鲁棒性低的缺点,本文提出了一种基于灰度渐变一致性的运动车辆阴影检测算法.首先应用改进的高斯混合模型对背景进行自适应重建和更新,然后根据差分图像中运动阴影在水平和竖直方向上灰度变化一致的特点,提取阴影区域的灰度跳变点,并以灰度跳变点的密度分布为依据分割车身区域和阴影区域,实现对阴影区域的识别与提取.实验结果表明,该算法能够快速有效地提取运动车辆的阴影,同时,本算法在阴影与相邻车辆车身重叠情况下也有较好的检测效果.

In view of the disadvantages that the robust of shadow detection algorithm based on shadow color feature is low,a moving vehicle＇s shadow detection algorithm based on gray gradual consistency is presented.First,using improved Gauss mixture model to reconstruction and update the background area.Then,according to the characteristics that the moving shadow which in the differential image has the similar gray change in horizontal and vertical direction,we could extract the shadow area＇s edge pixels.Finally,according to the density of these pixels,the shadow area and the vehicle area could be recognized.The experimental results show that the presented algorithm can effectively remove the shadow area.Meanwhile,even the shadow area and vehicle area overlap,this algorithm also has good detection effect.