摘要
考虑到人类视觉对图像轮廓特征的敏感性,将目标检测与轮廓提取结合起来,实现了目标轮廓自动提取的方法。首先采用了背景差法跟踪视频图像序列中的运动目标,并采用了自适应背景更新的方法更新背景图像,结合活动轮廓模型法GVFSnake进行目标轮廓的提取,从而得到具有精确边界的运动目标。实验结果表明这种方法运算速度快、能够快速地收敛到目标轮廓、准确地跟踪目标,实现实时自动轮廓跟踪。
Considering the sensitivity of the human vision about Contour Feature, this paper combined Object Detection with contour extraction. A new algorithm of automatic contour extraction method is realized. First, a tracking method which detected Moving Target using Background Difference from video sequences is adopted. Then, the background image is updated with the adaptive Background update method. For the accurate edge detecting reason, GVF （gradient vector flow） is applied to get the target contour feature. The experimental results indicate that this algorithm can track target accurately and automatically. The convergence speed is well.
出处
《微计算机信息》
北大核心
2008年第30期302-303,314,共3页
Control & Automation
关键词
目标检测
自适应背景更新
活动轮廓模型
Object detection
Adaptive background update
Activities Contour Model
