Research on Visibility Estimation Model Based on DenseNet

Abstract In recent years, the road visibility detection method based on video has been paid more and attention. It overcome deficiency of laser meter to some extent. Deep learning a good effect in image processing analysis. This paper firstly analyzes current situation deep learning, then compares DenseNet ResNet propose estimation model DenseNet. The integrates airport data data. Secondly, algorithm is used automatically extract features set. Finally, Softmax classifier constructed evaluate accuracy. They reduce problem disappearing gradient, enhance feature propagation, encourage functional reuse, greatly number parameters, well train model, effect. On this basis, Canny operator lane dividing line extraction edge analysis detection, do corresponding test. Kalman filter built given data, Gaussian process regression predict fog change trend.