基于虚拟现实的复杂场景非显著目标跟踪方法研究  

Research on non-salient target tracking method in complex scene based on virtual reality

作　　者：单盛[1] SHAN Sheng(Software College,Anhui Vocational College of Electronics&Information Technology,Anhui Bengbu 233000,China)

机构地区：[1]安徽电子信息职业技术学院软件工程学院,安徽蚌埠233000

出　　处：《齐齐哈尔大学学报（自然科学版）》2023年第1期53-58,共6页Journal of Qiqihar University(Natural Science Edition)

基　　金：安徽省高校自然科学研究项目“虚拟现实技术在影视动画中的实时场景交互应用研究”(KJ2021A1487)。

摘　　要：为避免复杂场景中的干扰因素影响非显著目标跟踪准确度,提出基于虚拟现实的复杂场景非显著目标跟踪方法。通过中值滤波设置图像窗口大小,统计相邻像素点最大中值,定义复杂场景特征区间。基于虚拟现实技术转换跟踪平面三维信息,利用正交原理计算空间位置,对应区间目标坐标。设定模糊逻辑集合,采用最大隶属度量方法构建隶属度函数,在复杂场景内跟踪非显著目标,完成非显著目标跟踪。设定多组目标跟踪场景,分别在影响因素较少、光感条件变化强烈、快速运动过程中和遮挡干扰背景五组条件下,测试非显著目标跟踪效果。实验结果表明,所提方法在上述设定中均可以完成准确跟踪,定位得到目标物。In order to avoid the interference factors in complex scenes affecting the tracking accuracy of non-significant targets, a virtual reality based non-significant target tracking method in complex scenes is proposed.Set the size of the image window through median filtering, count the maximum median of adjacent pixels, and define the feature interval of complex scene. Based on virtual reality technology, the three-dimensional information of tracking plane is converted, and the spatial position and the corresponding interval target coordinates are calculated by using the orthogonal principle. The fuzzy logic set is set, and the membership function is constructed by using the maximum membership measure method to track non-significant targets in complex scenes and complete the tracking of non-significant targets. Multiple sets of target tracking scenarios were set up to test the tracking effect of non-significant targets under five groups of conditions, namely, less influencing factors, strong changes in light perception conditions, rapid movement and occlusion interference background. The experimental results show that the proposed method can accurately track and locate the target in the above settings.

关 键 词：虚拟现实 复杂场景 非显著目标 目标跟踪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]