涉电公共安全场景实时跟踪和防抖算法研究  

Real-Time Planar Tracking and Jitter Reduction in Electricity-Related Public Safety Environments

作　　者：杨子力 陈武 朱家山 李胤廷 

机构地区：[1]云南电网有限责任公司曲靖供电局,云南 曲靖

出　　处：《计算机科学与应用》2020年第7期1414-1421,共8页Computer Science and Application

摘　　要：针对涉电安全工作与治理工作的增强现实展示应用,需要在移动平台上实现快速平面识别和位置姿态估算,并且叠加虚拟场景与实际平面联动的需求,提出一种快速而高效的三维平面物体识别和跟踪算法。同时,为了解决物体识别和跟踪过程中的数据抖动问题,采用了Butterworth滤波器构建了位置和姿态数据的后处理平滑功能模块,以极低的系统延迟为代价,将输出的结果数据进行了平滑和防抖动处理。实验表明,该方法可以快速、低延迟、无抖动地将虚拟电站培训场景与摄像头呈现的现实画面叠加渲染,实现虚实结合的展示效果,从而满足了实际使用中的需要。This paper combines the specific needs of fast plane recognition and poses estimation in augmented reality applications, to achieve a fast and efficient three-dimensional planar object identification and tracking algorithm. In order to solve the jitter problem during tracking, this paper also implements a post-processing module for position and attitude data smoothing, based on the Butter-worth filter, which handles data at the expense of very low system latency, to satisfy the needs of practical uses in virtual power station training scenes to achieve a combination of virtual and reality display.

关 键 词：增强现实 平面跟踪 防抖动 滤波 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]