摘要
交通图像分析是智能交通领域的关键技术之一。为实现复杂交通场景中的多目标检测与跟踪,设计了一种结合小波提升框架和KLT特征点跟踪的多运动目标检测与跟踪算法。对序列图像中相邻两帧图像的融合图像进行小波提升变换,求取水平和垂直方向上的小波能量,通过合理阈值二值化小波能量矩阵,再利用贴标签方法检测出运动目标;利用KLT特征点集合代表目标,通过跟踪后的特征点集合与目标检测区域的相互关联,实现多目标的跟踪。实验结果表明了所提算法的有效性。
Video analysis is one of the key techniques in ITS. A method combining wavelet lifting scheme and KLT feature point tracker is designed for multiple object detection and tracking. The wavelet lifting scheme is applied to the two continuous frames to gain the vertical and horizontal wavelet energy map. Moving object is then detected by thresholding the energy map. Represented by KLT feature points groups, multiple objects are tracked through the association between feature groups and wavelet detection blobs. Experimental results validate the proposed method.
出处
《交通信息与安全》
2011年第5期135-139,共5页
Journal of Transport Information and Safety
关键词
交通
图像处理
小波
目标检测
目标跟踪
transportation
image processing
wavelet
object detection
object tracking
