多摄像机协同环境下的目标轨迹标识
MOTION DETECTION IN MULTI-CAMERA COLLABORATION SYSTEM
针对多摄像机协同问题,提出基于对极几何和三焦点张量的目标匹配方法。利用对极几何对两个摄像机之间的目标进行匹配。然后利用三视图之间的三焦点张量得到第三视图的对应转移点并确定目标位置。此方法有不依赖景物内容、无需标定摄像机的优势。本文建立了多摄像机协同系统,实现对三视图同一目标的匹配和轨迹标识,对比了三焦点张量与对极线交点的点转移结果。事实证明三焦点张量有着更好的性能,避免了对极线交点退化问题的出现。
his paper presents a multi-cameras collaboration methods based on epipolar geometry and trifocal tensor.In this method,we use epipolar to match targets in two camera views,and then use trifocal tensor to implement point transfer and locate targets in three camera views.Comparing with other methods, the proposed one doesn't need camera calibration or content of the scene.A multi-cameras collaboration and target identification system is built in this paper using three views by epipolar point transfer and tensor point transfer.The experimental result shows trifocal tensor point transfer has a better performance,which could avoid the occurrence of the degradation of epipolar intersections..
苏菲、王璐
摄影技术电子技术应用计算技术、计算机技术
多摄像机协同对极约束三焦点张量
Multi-camera collaborationEpipolar geometryrifocal tensor
苏菲,王璐.多摄像机协同环境下的目标轨迹标识[EB/OL].(2014-07-08)[2025-08-02].http://www.paper.edu.cn/releasepaper/content/201407-101.点此复制
评论