【Key words】 Difference picture； Face detection； Face recognition； Target tracking； Motion detection；
基于6个网页-相关网页
In this paper, a difference image based motion detection algorithm, which results in the symbolic representations of motion targets via their outer rectangular bounds, is presented. Then according to the continuity constraint, motion targets are tracked through image sequence.
文中介绍了一种基于差分图象的运动目标检测算法，检测结果是符号化了的图象，其中运动目标由其外接矩形表示，然后根据连续性约束假设，实现了运动目标的跟踪。
参考来源 - 基于差分图象的多运动目标的检测与跟踪For a video sequence, moving change regions are achieved firstly by means of converting difference image to binary image using adaptive thresholding and morphological filtering. Then, with the edge information of current frame, the edges of moving objects are generated.
首先通过帧间差分和自适应阈值得到二值差分图象，经形态滤波提取出运动变化区域，然后结合当前帧的边界信息确定运动目标的边界，最后由区域填充得到连通的运动目标区域并检测出运动目标。
参考来源 - 视频序列中运动目标的检测与跟踪·2,447,543篇论文数据，部分数据来源于NoteExpress
本文的主要研究贡献有：1。提出基于CNN差分图象合并的视频分割算法。
The main novel contributions of this paper are as follow: 1. Difference merged image algorithm based on CNN is presented.
不同差分方法下的实验结果表明，这种基于差分图象的运动目标检测框架是行之有效的。
A great deal of experiments in the case of different difference show that this framework is effective.
针对背景相对静止的视频序列，提出了基于CNN差分图象合并的视频分割算法，并构建了与该算法相关的五个CNN模板。
Aim at video sequences with static background, the difference merged image algorithm based on CNN is presented. In order to realize the algorithm, five CNN templates are constructed.
应用推荐