【作者】 骆玉荣； 刘建丽； 史晓涛；

【Author】 LUO Yu-rong LIU Jian-li SHI Xiao-tao College of Computer Science and Technology, Beijing University of Technology, Beijing, 100022,China

【机构】 北京工业大学计算机学院；

【摘要】 对于自动获取的车辆图像,论文中提出了一种自动车窗识别方法。由于车窗位置特征的不明显性,针对不同光照条件下、具有复杂背景的深色车和浅色车分别进行处理。在识别深色车车窗部分,基于Log边缘检测后水平投影图确定车辆垂直方向的中心位置,然后以定位好的车牌位置为参照,利用滑动窗口定位车窗。在识别浅色车车窗部分,提出了一种直线拟合确定车窗位置的方法。实验结果显示,自动车窗识别方法对实验数据可得到90%以上的正确率。更多还原

【Abstract】 To vehicle images captured automatically, a method about automatic identification of vehicle windows is presented in the paper. Due to the unobvious characteristic of vehicle windows’ place, it deals with the light-colored vehicles and dark-colored vehicles with different light conditions and complicate backgrounds respectively. In the identification of dark-colored vehicles’ window, a horizontal projection histogram by the Edge Detection Based on Log is applied to determine the center location of the vehicles in vertical direction, and the sliding window is used to locate the vehicle window, taking the vehicle plate which has been located as a reference. In the identification of light-colored vehicles’ window, a linear fitting method is presented. Experiment result reveals that the method can locate the vehicles window accurately and can reach 90% accuracy.更多还原