一种面向城市道路的视觉图像去雾方法

A Visual Image Dehazing Method for Urban Roads

针对雾天环境的城市道路下自动驾驶车辆视觉感知效果不佳的问题,提出一种基于大气光值的快速图像去雾改进算法,并验证了算法的有效性。首先,制定了大气光动态估算策略,并设计了大气光动态估算的自适应触发函数,通过大气散射模型进行了大气光的估计;其次,利用最小滤波技术获取了雾天图像的暗通道图,并估算了对应的投射图像;再次,制定了大气光计算策略,并优化了去雾系数;最后,利用直方图均衡化算法抑制了残余噪声,进一步提升了无雾图像的对比度。实验结果表明,所提算法相比DCP、CAP等在NIQE和SSEQ性能指标上都有所提升,拥有更好的细节恢复能力和处理性能,更有利于交通信息的提取。

Aiming at the problems in visual environment perception of autonomous vehicles in the foggy environment of urban roads,an improved fast dehazing algorithm based on atmospheric scattering model and image processing technology is proposed,and the effectiveness of the algorithm is verified by experiments.Firstly,the atmospheric light dynamic estimation strategy is formulated,the adaptive trigger function for atmospheric light dynamic estimation is designed,and the atmospheric light is estimated by the atmospheric scattering model.Secondly,the dark channel map of the hazy image is obtained by the minimal filtering technique,and the transmission map is estimated.Thirdly,the atmospheric light calculation strategy is formulated,and the dehazing coefficient is optimized.Finally,the residual noise is suppressed by the histogram equalization algorithm,which further improves the contrast of the haze-free image.The experimental results show that,compared with algorithms such as DCP and CAP,this algorithm has improvement in NIQE and SSEQ performance indicators,with better detail recovery ability and processing performance,and is more conducive to the extraction of traffic information.