摘要： 提出一种基于边缘特征的阴影消除方法，通过边缘检测获取阴影边缘信息，利用边缘差分、形态学等运算进行阴影消除。并提出一种基于灰度特征的阴影消除方法，利用暗化因子高斯模型进行阴影消除。结合2种方法的优点，给出一种基于多特征融合的运动车辆阴影消除方法，可同时解决车辆与阴影颜色相似以及阴影内部边缘复杂等原因造成的误检问题。实验结果表明，该方法具有较好的实时性、精确性和鲁棒性。

关键词: 多特征融合, 边缘检测, 高斯模型, 阴影消除, 智能交通

Abstract: This paper proposes a shadow removal method based on edge feature. By using the edge information of the shadow which is obtained by edge detection algorithm the shadow of moving vehicle is removed by edge difference analysis and morphological operations. Then, a shadow removal method based on gray feature is proposed, which remove the shadow of vehicle using Gaussian darkening factor model. Combined with the advantages of two methods, a method of shadow removal for moving vehicle based on multiple feature fusion is proposed. It can effectively solve the problem of wrong detection caused by complex edges in shadow region and the similar color between vehicle and shadow. The testing results demonstrate that by using the proposed method shadow removal can be realized with real-time performance, strong robustness and high accuracy.

Key words: multiple feature fusion, edge detection, Gaussian model, shadow removal, intelligent transportation

中图分类号: 

• TN911.73