AlTwo: Vehicle Recognition in Foggy Weather Based on Two-Step Recognition Algorithm.

This paper focuses on the application of vehicle recognition in foggy weather, and proposes a two-step recognition algorithm based on deep learning, hoping to still have a good recognition result under the influence of fog. In AlOne, we use the current popular defogging algorithms: adaptive histogram equalization, single-scale Retinex, dark channel priori for image defogging. In AlTwo, we build a convolutional neural network model based on AlexNet to recognize vehicle image and obtain prediction accuracy. By comparing different performance indicators, the best performing dark channel prior algorithm is selected as the defogging algorithm. In the end, we improved the accuracy of vehicle recognition under the influence of low, medium and high fog concentrations to more than 97%.