摘要
提出了一种自适应的阴影检测方法,去除了传统固定阈值阴影检测方法残留的阴影边缘,有效地改善了阴影检测效果。首先采用kmeans聚类、求前景灰度直方图峰值间平均值等方法得到自适应的阈值,在此基础上,计算满足阈值约束的前景像素点,将该点及其8邻域点都作为可去除的阴影点进行标记。最后,去除标记的阴影点及极小面积的前景区域。本文对已有的阴影检测算法进行了改进,加入了自适应的阈值计算方法并去除了原有算法检测后残留的阴影边缘,在对室内及室外视频序列进行的检测中都取得了较好的效果。
This paper presents an adaptive way for shadow detection. The residual shadow edge appearing in fixed thresholds method is eliminated in our algorithm and the detection performance is better. We first get adaptive thresholds using kmeans clustering,averaging foreground grey level histogram peak value,etc. Based on the adaptive thresholds,foreground pixels satisfying constraints are calculated. These pixels and their 8 connected pixels are marked as shadow pixels. Finally,we remove all the shadow pixels and small area foreground. In this paper,we propose an improvement to an existing shadow detection algorithm.Adaptive method is added and shadow edge is wiped out. The algorithm works well for both indoor and outdoor sequences.
出处
《信号处理》
CSCD
北大核心
2014年第11期1370-1374,共5页
Journal of Signal Processing
基金
科技部973项目(2012CB316304)
国家自然科学基金(61172128)
