一种基于局部加速的实时精确虚实融合方法

An accurate real-time virtual reality fusion method based on local acceleration

在虚实融合中,虚实场景显示的较高精确性和较好实时性是良好用户体验的必要条件。然而,环境干扰、算法时间空间复杂度以及高帧频处理速度等问题严重影响虚实融合效果,现存的虚实融合方法很难同时满足精确性和实时性的要求。针对此难点问题,提出了一种基于局部加速的精确虚实融合方法,利用明显特征建立虚实场景映射关系,同时对融合区域进行局部搜索匹配。具体创新体现在以下2点:1)提出了一种虚实场景的普适标定配准方法;2)提出了一种待融合目标的局部搜索加速策略。实验证明,提出的方法具备较好的鲁棒性并达到高帧频处理速度,已成功应用在电力培训仿真系统中。

In the fusion of virtual and actual reality, the high accuracy and strong real-time performance of virtual and real scenes are the necessary conditions for the good user experience. However, the problems such as the environment interference, the time and space complexity of the algorithm, and the running speed of high frame rate, etc., affect the effect of virtual-real fusion. The existing virtual-real fusion methods are difficult to meet the requirements of both accuracy and real-time simultaneously. Aiming at the difficult problem, this paper presents an accurate virtual-real fusion method based on local acceleration. The virtual and real scene mapping relationship is established by using obvious features, and the local search and match are carried out for the fusion area at the same time. The specific innovations are as follows: 1) a general calibration registration method is proposed for virtual and reality scenes;2) an effective local search acceleration strategy is proposed for the target to be fused. Experiment results show that the proposed method has good robustness and high frame frequency processing speed, and it has been successfully applied in the electric power training simulation systems.