为解决当前夜间辅助驾驶仪平面显示导致驾驶员无法正确判断景物纵深关系的问题，本文提出利用平行式红外双目立体系统进行夜间辅助驾驶的方案。针对红外探测器分辨精度低导致观看红外立体影像时立体盲区占立体区比重过大的问题，本文通过分析对比平行式红外双目系统和肉眼的立体感知精度，结合 Panum融合区理论，对红外立体影像立体区间进行界定，并对立体区间所对应的基线长度范围进行推导。根据推导结果，计算分析 3组不同场景下基线长度所满足的最小值，并以得到的最小值为基准获取 5组不用基线长度下的视差图。最后通过虚拟现实显示装置验证了推导结果的正确性。该结果可用于指导红外双目系统参数的选取和搭建。

To solve the problem where the driver cannot judge the depth relationship of the scene from the two-dimensional display, this study proposes the use of a parallel infrared binocular stereo system for night driving. Considering the low resolution of infrared detectors, it is observed that the stereo blind area accounts for a large proportion of the stereoscopic area when viewing the infrared stereo images. Furthermore, the minimum resolution distance of the infrared binocular system and human eye stereoscopic perception is analyzed. Combining the theory of Panum’s fusional area, the stereo interval of infrared stereo image is defined, and the range of baseline length corresponding to the stereo range is deduced. According to the results of the derivation, the minimum value of the baseline length is calculated under three groups without using the minimum value, and the five parallax images without the baseline length are obtained on the basis of the minimum value. Finally, the derivation result is verified by the virtual reality display device. The results can be used to guide the parameter selection and construction of infrared binocular systems.