The vision problem of drivers during the night is mainly owing to the high-intensity headlight-beam of the oncoming vehicle from the reverse direction. It causes temporary blindness to the driver. To overcome this situation, people generally use color windshield glass, sun visor, and night-vision glass. But, these are not considered as the best solution since it decreases the light intensity of the entire view including the road. Many researchers used various image enhancement techniques to overcome this situation. But, the existing approaches are unable to dim the high-beam headlights of oncoming vehicles without affecting the road view. In this paper, a novel night-vision system is proposed to resolve the problem in real-time for manual-driving vehicles and autonomous vehicles. The proposed method includes region segmentation of frames, local enhancement techniques in different regions followed by adaptive Gaussian filtering. Pixels masking, gamma correction, and low-light pixel enhancement are applied to three distinct regions. Both autonomous vehicles and manual drivers can get a bright and a prominent view of the road with dim headlights of oncoming vehicles in real-time. Numerous heuristic real-time test reveals the performance superiority of the projected system compared to state-of-art methods in quantitative as well as qualitative point of view.

夜间驾驶员的视力问题主要归因于迎面驶来的车辆从反方向发出的高强度大灯束。这会导致驾驶员暂时失明。为了克服这种情况，人们通常使用彩色挡风玻璃，遮阳板和夜视玻璃。但是，这些方法不被视为最佳解决方案，因为它会降低包括道路在内的整个视图的光强度。许多研究人员使用各种图像增强技术来克服这种情况。但是，现有方法无法在不影响道路视野的情况下调暗迎面驶来的车辆的远光灯。本文提出了一种新颖的夜视系统，以实时解决手动驾驶车辆和自动驾驶车辆的问题。所提出的方法包括帧的区域分割，局部增强技术在不同区域，然后进行自适应高斯滤波。像素遮罩，伽玛校正和低光像素增强被应用于三个不同的区域。无人驾驶汽车和手动驾驶者都可以通过近来车辆的前大灯实时看到明亮明亮的道路景观。在定量和定性的观点上，大量的启发式实时测试显示了与最新方法相比，投影系统的性能优势。无人驾驶汽车和手动驾驶者都可以通过近来车辆的前大灯实时看到明亮明亮的道路景观。在定量和定性的观点上，大量的启发式实时测试显示了与最新方法相比，投影系统的性能优势。无人驾驶汽车和手动驾驶者都可以通过近来车辆的前大灯实时看到明亮明亮的道路景观。在定量和定性的观点上，大量的启发式实时测试显示了与最新方法相比，投影系统的性能优势。