In addressing the challenge of detecting lane markings in foggy conditions, a novel multi-lane line detection method was developed utilizing the LaneNet algorithm alongside image enhancement techniques. This approach aims to improve the precision of lane line identification by neural networks under adverse weather circumstances. The images affected by fog were processed using both the dark channel dehazing technique and the multi-scale Retinex method before being fed into the neural network. Additionally, an attention mechanism known as CBAM was integrated into the feature extraction component of LaneNet, while an asymmetric convolution was introduced to account for the elongated nature of lane lines, thereby enhancing the network's ability to extract relevant features from images and capturing more detailed information. Experimental findings reveal that this enhanced version of LaneNet achieved a 2.67% improvement in lane line detection accuracy on the Tusimple dataset during foggy conditions when compared to its original counterpart.