摘要
对双通道视频图像的多特征并行进行融合,能够有效提高双通道图像的完整性。对视频图像多特征并行融合,需要获得运动图像软特征,跟踪当前帧位置,完成视频图像多特征并行的融合。传统方法首先融合视频图像多特征并行概要,对其进行模板识别匹配,但忽略了跟踪当前帧位置,导致融合效果不理想。提出双通道视频图像多特征并行融合方法。融合待跟踪目标区域,统计初始像素点拟合成时域轨迹,计算时域轨迹的空间谱带以及边缘谱带进行合并,获得运动图像软特征,估计出单应性矩阵,跟踪当前帧位置。实验结果表明,该方法有效提高了双通道图像的完整性。
The aim of this article is to overcome defect of traditional method of multi - feature parallel fusion of video image with dual channel, such as not ideal fusion effect. A new method of the multi - feature parallel fusion is presented. Firstly, integrated with target area to track, initial pixel point is counted up and is fitted to time domain track. Then, spatial spectrum band and marginal spectrum band of the track are calculated. Soft feature of motion image is obtained and homograph matrix is estimated. Current frame location is traced. Simulation results reveal that the method can improve integrality of the image with dual channel effectively.
作者
黄艺坤
HUANG Yi - kun(Concord University College, Fujian Normal University, Fuzhou, Fujian, 350117, China)
出处
《计算机仿真》
北大核心
2018年第4期154-157,共4页
Computer Simulation
基金
福建省教育厅科技项目:移动增强现实技术在室内定位和导航的研究(JAT160670)
2017年福建省中青年教师教育科研项目:基于图像识别和物联网技术的车位空位引导系统的研究(JZ170367)
关键词
双通道
视频图像
多特征并行融合
Dual channel
Video image
Multi - feature parallel fusion
