Abstract: The large-scale topological structures of the object are first perceived in the visual perception process.When the motion state of the scene object changes, the large-scale topological relationship of its space also changes.In the visual cognition process, when the aforementioned two large-scale phenomena appear simultaneously, determining how the cognitive priority of human vision with moving objects occurs is an important basic problem.Therefore, based on virtual reality and eye-tracking technology, this study proposes a virtual experimental method to further explore the cognitive characteristics of human vision on object motion cognition, in which two important concepts are employed including the topology complexity and spatial topological relationship.Based on this, an experiment is conducted to use the response surface analysis method to design the random experimental values of independent variable factors in the scene, and the visual method is used to compare the results of eye movement data analysis with non-inductive feedback in multiple groups of contrast scenes.The results of the new method demonstrate that 516 of the 545 eye-tracking data collected by a Tobii eye tracker occur on moving objects.That is, compared with the large-scale topological property changes of the object's attributes, the human visual perception process prioritizes identifying the large-scale spatial topological relationship changes of object motion.The motion cognitive priority provides new research insight into the visual gaze prediction technology of credible artificial intelligence driving.In addition, the validity of the proposed experimental analysis method is verified.

Key words: human eye vision, topological perception theory, spatial topological relationship, motion cognition, eye tracking trajectory, virtual driving modeling

摘要： 视觉感知过程源于对象自身的大范围拓扑性质，而场景对象发生运动状态变化时，其所在空间的大范围拓扑关系也将随之改变。为探索大范围拓扑性质和运动性质同时改变时人眼视觉的认知优先特性，基于虚拟现实技术及眼动跟踪技术，通过构建虚拟驾驶及眼动轨迹分析实验平台，结合拓扑复杂度和空间拓扑关系概念分别探讨二维和三维空间中人眼视知觉对物体运动的认知特性。在此基础上，采用响应面分析方法设计场景中自变量因子的随机实验数值，通过可视化方法对比多组认知场景的无感反馈眼动监测结果。实验结果表明，在Tobii眼动仪采集的545个眼动数据中，发生在运动物体上的数据有516个，相比于物体自身属性的大范围拓扑性质变化，人眼视知觉过程更优先识别物体运动的大范围空间拓扑关系变化。该运动认知优先特性为可靠人工智能驾驶中的视觉注视预测技术提供了新的研究思路，同时验证了所提实验分析方法的有效性。

关键词: 人眼视觉, 拓扑知觉理论, 空间拓扑关系, 运动认知, 眼动跟踪轨迹, 虚拟驾驶建模

CLC Number: 

• TP391