摘要
针对Snake不能收敛于凹形边缘和收敛速度慢的缺点,结合帧间差分和梯度矢量流主动轮廓模型(GVF Snake),提出了一种运动目标跟踪方法。首先通过帧间差分得到不同时刻的灰度差异图像,然后用形态方法得到的运动目标粗略轮廓作为Snake的初始轮廓,最后利用GVF Snake的特性,准确地收敛到目标边界。实验证明该方法运算速度快、能够快速地收敛到目标轮廓、准确地跟踪目标,并具有一定的抗噪能力。
Aiming at the limitations that classic Snake cannot converge to concave edge and convergence speed is slowly, a moving target tracking method based on frame-to-frame difference and GVF snake is proposed. First, the map of gray difference is gained by frame-to-frame difference algorithm. Then coarse contour is obtained using morphological operator and taken as initialized contour of GVF snake. Finally, GVF Snake evolves to precise target contour. The method can converge to target contour rapidly. Experiments show that it can track moving target fast and exactly, and it is robust to noise.
出处
《红外与激光工程》
EI
CSCD
北大核心
2007年第2期226-228,256,共4页
Infrared and Laser Engineering
基金
国防科工委计量技术基础项目(J182001C002)
关键词
帧间差
GVF
SNAKE
形态学算子
运动目标跟踪
Frame-to-frame difference
GVF Snake
Morphological operator
Moving target tracking
