基于视频图像的汽车乘员位置追踪算法研究  

Study on Vehicle Occupant Position Tracking Algorithm Based on the Video Images

作　　者：曹立波[1] 刘曜[1] 周雪桂 

机构地区：[1]湖南大学汽车车身先进设计制造国家重点实验室,长沙410082 [2]长沙立中汽车设计开发有限公司,长沙410205

出　　处：《中国机械工程》2013年第8期1126-1130,共5页China Mechanical Engineering

基　　金：长沙市科技计划资助项目(K1101076-11);湖南大学汽车车身先进设计制造国家重点实验室自主三课题资助项目(61075004)

摘　　要：采用基于视频图像的目标追踪算法检测副驾驶员座位上的乘员位置,并判断其是否处于离位状态。该算法运用背景差分法并辅以基于肤色信息的人脸识别算法将乘员从图像中分离出来,然后通过Canny算子提取边缘信息,并利用椭圆拟合出乘员上半身轮廓,根据椭圆方程判断其离位状态。实验结果表明,该算法对实现视频序列图像中乘员位置的检测和连续追踪是可行的。An object tracking algorithm based on video images was used to detect the occupant po- sition on the co--driver seat and to judge the in position or out of position state. The background sub- traction and skin color based face recognition algorithm were applied to segment the occupant from the images. Then, edge information was extracted with Canny operator and upper silhouette was fitted with an ellipse. The in or out of position state was judged through the ellipse equation. The experimen_ tal results show that using the algorithm to detect and track the occupant position in the sequent video images is feasible.

关 键 词：视频图像 乘员轮廓 位置追踪 椭圆拟合 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]