摘要
首先设计了一种新的预处理流程,去除非阴影及车辆区域的边缘;其次,对边缘图像进行填充,得到运动部分的轮廓边缘图;最后,建立边缘擦除法则,擦去阴影部分的边缘,并对边缘图进行填充,得到最终的去除了阴影的目标图像.算法的核心是将前景图像中的阴影边缘从目标边缘上分离出来,着重解决了实际应用中经常出现的目标轮廓图边缘和实际阴影轮廓边缘不重合造成阴影边缘无法去除的问题,同时很好解决了多个前景目标因阴影而粘连的情况.大量实际道路视频图像的测试表明,本算法去除阴影效果好,有较强的实际应用价值.
The new algorithm is ,firstly ,related to the designation of a new pre‐process method to remove the un‐shadow and un‐vehicle＇s edge ,secondly ,to the fulfillment of the edge image to get the foreground＇s contour edge of moving part ,and ,lastly ,to establishment of edge erase rules to remove the shadow ed‐ges ,and to fill the marginal figure ,in order to get the final target image that its shadow has been re‐moved .The kernel of this algorithm separates the shadow edge from the foreground .In practical applica‐tions ,the foreground＇s contour edge is often not coincide with shadow＇s contour edge ,so it will be difficult to remove the shadow edges ,this algorithm can solve this problem .When the multiple vehicles adjoined with each other by shadows ,this method can also split them .Experiments on practical traffic video se‐quences demonstrate that our method is efficient and robust ,it has strong practical application value .
出处
《西南师范大学学报（自然科学版）》
CAS
CSCD
北大核心
2014年第11期90-95,共6页
Journal of Southwest China Normal University(Natural Science Edition)
基金
广西教育厅科研项目(201106LX652)
广西民族师范学院科研项目(XYYB2010005)
关键词
运动阴影去除
边缘特征
车辆检测
背景差法
moving shadow removal
edge feature
vehicle detection
background subtraction
