The invention discloses an expressway foggy weather visibility detection method based on dark channel prior and deep learning. The method comprises the following steps: 1, acquiring a video image acquired by an expressway camera, and building a background model; 2, acquiring an initial transmittance graph: 21, building a six-layer convolutional neural network, and performing training by using a large quantity of samples to obtain a network model; 22, loading a convolutional neural network model, and initializing the network by using the length and width of an original picture; 23, transmitting an original foggy weather image into the network for computing to obtain a corresponding transmittance graph; 3, optimizing the transmittance graph; 4, solving an average atmospheric extinction coefficient: 41, specifying at least two lane white lines in an original image to select road key points; 42, acquiring transmittance at two ends of each lane line, and computing corresponding atmospheric extinction coefficients respectively; 43, solving an average atmospheric extinction coefficient; and 5, estimating the visibility.