摘要
文章提出一种光流场和Grabcut算法融合的运动目标分割方法,利用光流场计算相邻两帧图像得到运动目标初步检测的模糊结果,将结果处理后作为Grabcut前景标定依据,取代Grabcut人工交互过程,得到最终运动目标精确检测结果。
The aim of this paper is to propose a method for the segmentation of moving object based on a combination of optical flow field and Grabut algorithm. The optical flow field is used to calculate two adjacent pictures to obtain blurry result of initial detection of the moving target. Instead of the manual interaction process of Grab Cut, the result will be treated as basis for foreground calibration of Grab Cut after being processed. A final accurate detection result of the moving target is obtained.
出处
《电子技术（上海）》
2015年第2期3-6,共4页
Electronic Technology
