摘要
针对普通基于HSV颜色空间算法的缺陷,提出了一种新颖的车辆阴影消除方法.通过自适应混合高斯模型(AGMM)进行背景建模得到视频序列背景和前景,使用改进的高斯-拉普拉斯(LOG)边缘检测算子对前景图像和阈值分割后的前景进行两次边缘检测,最后对两次检测后的前景边缘进行相减,保留车辆运动区域.实验结果表明,提出的新颖算法能有效去除与车辆特性相近的光照阴影.
Since outdoor sports vehicle shadows exist in the process of vehicle detection,this paper proposes a novel method of vehicle shadow elimination.With the adaptive Gaussian mixture model (AGMM) for background modeling,we can get the background and the frame of the video sequence,we can use the improved Laplace of Gaussian function (LOG) edge detection operator to conduct two times edge detection to the foreground image and the foreground of threshold segmentation.Finally,for keeping the vehicle movement area,we should have the foreground edges of the two times detecting results subtracted.This experimental result shows that the novel algorithm can effectively remove the light shadow which is close to the vehicle characteristics.
出处
《郑州大学学报(工学版)》
CAS
北大核心
2014年第5期11-14,共4页
Journal of Zhengzhou University(Engineering Science)
基金
国家自然科学基金资助项目(61306106)