摘要
提出了一种新的视频运动目标检测与跟踪方法。该方法首先采用自适应帧间差分法对视频序列图像的运动目标进行粗检测,进而采用BVF(边界矢量场)Snake方法准确地检测出运动目标轮廓;其次获得轮廓质心后,对传统的α-β-γ滤波器进行了改进,实现对运动目标的实时跟踪;最后根据预测的质心位置来自动完成下一帧轮廓初始化。实验结果表明,该方法具有良好的实时性和准确性。
Anew method of moving object detection and tracking in video sequences was proposed. Firstly, the adaptive frame-to-frame difference method was used to complete the coarse detection for moving object in video sequence images, and the method of Boundary Vector Field （BVF） Snake was used to detect the contours exactly. Secondly, the traditional method of α-β-γ was improved to achieve real-time tracking for the moving object after obtaining the centroid of the contours. The contours of the next frame was initialized automatically according to the predicted centroid finally. Experimental results show that the proposed method is well real-time and has high accuracy.
出处
《计算机应用》
CSCD
北大核心
2008年第B06期223-225,229,共4页
journal of Computer Applications
关键词
帧间差
BVF
Α-Β-Γ滤波
目标检测与跟踪
frame-to-frame difference
Boundary Vector Field （BVF）
α-β-γ filter
object detection and tracking
作者简介
林乃昌（1982-），男，福建长汀人，硕士研究生，主要研究方向：图像处理、模式识别；（nchlin@126．com）
胡步发（1963-），男，福建宁德人，副教授，博士，主要研究方向：计算机视觉、模式识别。
