重载铁路车站裸眼全息三维成像的实时空间映射方法研究

Real time spatial mapping method for naked eye holographic 3D imaging of heavy-duty railway stations

摘要:

针对重载铁路车站裸眼全息三维成像中因环境噪声干扰,导致在低能见度工况下出现成像失效问题,无法确保关键细节,造成强光下梯度反转现象,影响实时空间映射效果。为此,提出一种融合多尺度权重计算与亮度-梯度协同引导的实时空间映射方法。构建拉普拉斯对比度权值、显著性权值、空间信息权值的协同优化模型,通过引导滤波器平滑权值分布,解决重载场景中局部模糊与噪声敏感问题。引入实时空间映射权重优化亮度层,避免低能见度工况下出现成像失效问题,保护高反光区与暗区细节。构造权重加权的梯度引导项,抑制粉尘干扰导致的梯度失真,避免强光下的梯度反转现象。建立图像点至空间平面的多项式映射关系,适配列车、起重机等异形目标,实现亚米级定位精度。裸眼三维感知实验结果表明,该方法在煤尘干扰、设备振动等复杂场景下,几何畸变率降至0.38,视觉舒适度显著提升,无重影眩晕现象,满足车站实时监测的精度与响应需求。

Abstract:

In response to the problem of imaging failure in low visibility conditions caused by environmental noise interference in naked eye holographic 3D imaging of heavy-duty railway stations, key details cannot be guaranteed, resulting in gradient reversal under strong light and affecting the real-time spatial mapping effect. To this end, a real-time spatial mapping method is proposed that integrates multi-scale weight calculation with brightness gradient collaborative guidance. Build a collaborative optimization model for Laplacian contrast weight, saliency weight, and spatial information weight, and solve local blur and noise sensitivity problems in overloaded scenes by guiding the filter to smooth the weight distribution. Introducing real-time spatial mapping weight optimization brightness layer to avoid imaging failure in low visibility conditions and protect details in high reflective and dark areas. Construct a weighted gradient guidance term to suppress gradient distortion caused by dust interference and avoid gradient reversal under strong light. Establish a polynomial mapping relationship between image points and spatial planes, adapt to irregular targets such as trains and cranes, and achieve sub meter level positioning accuracy. The results of naked eye 3D perception experiments show that this method reduces the geometric distortion rate to 0.38 in complex scenarios such as coal dust interference and equipment vibration, significantly improves visual comfort, and eliminates ghosting and dizziness, meeting the accuracy and response requirements of real-time monitoring at stations.