For the future open 5G Internet of Vehicles (IoV), due to the flexibility and load sharing, the popular network architecture of IoV proposed by many studies is the mobile-edge computing (MEC) framework combining with software-defined networking (SDN). However, under this architecture, moving vehicles and MEC devices are not like the cloud SDN with strong security protection. Thus, identity verification is an important security issue. We find that if the identity credentials of vehicles and infrastructures are obtained by adversaries (i.e., identity theft), the current cryptography-based authentication methods cannot cope with this problem. In this article, we propose a behavior-based verification method, named Crowd-Learning, by utilizing the idea of crowd in software-defined vehicular networks with a MEC framework. In Crowd-Learning, we design an incentive mechanism to stimulate some MEC infrastructures to provide accurate and appropriate amount of data for future correct behavior estimation. Without knowing the model of the dynamic environment, this incentive mechanism needs to apply reinforcement learning to let MEC infrastructures learn how to send data based on the current state. Our Crowd-Learning method verifies vehicles and reduces the verification latency by estimating the vehicle’s behavior in advance. Meanwhile, it verifies infrastructures during the process of reinforcement learning based on the idea of crowd intelligence. The fake infrastructures and anomalous vehicles expose themselves when learning. In experiments, we use the traffic simulation tool, called simulation of urban mobility (SUMO), to generate extensive vehicle traces and evaluate the performance of the Crowd-Learning verification method. The results show that the Crowd-Learning verification method can ensure high verification accuracy for vehicles and infrastructures with satisfying low verification latency.

Capturing document images with hand-held devices in unstructured environments is a common practice nowadays. However, "casual" photos of documents are usually unsuitable for automatic information extraction, mainly due to physical distortion of the document paper, as well as various camera positions and illumination conditions. In this work, we propose DewarpNet, a deep-learning approach for document image unwarping from a single image. Our insight is that the 3D geometry of the document not only determines the warping of its texture but also causes the illumination effects. Therefore, our novelty resides on the explicit modeling of 3D shape for document paper in an end-to-end pipeline. Also, we contribute the largest and most comprehensive dataset for document image unwarping to date - Doc3D. This dataset features multiple ground-truth annotations, including 3D shape, surface normals, UV map, albedo image, etc. Training with Doc3D, we demonstrate state-of-the-art performance for DewarpNet with extensive qualitative and quantitative evaluations. Our network also significantly improves OCR performance on captured document images, decreasing character error rate by 42% on average. Both the code and the dataset are released.

Rail surface defects threaten train and passenger safety. Hence rail surfaces must be restored using different processes depending on measurement of the severity of the defects. In this paper, we propose a new method for automatic classification of rail surface defect severity from images collected by rail inspection vehicles. It contains 2 components: a rail surface segmentation module, which utilizes structured random forests to generate an edge map and a Generalized Hough Transform to locate the boundaries of the rail surface; and a defect severity classification module, which combines multiple classifiers through a stacked ensemble model. The first-level learners are trained using descriptors of the rail surface images extracted by texton forests and texton dictionaries, with x2-kernel SVM classifiers. The probability estimation output of the first-level learners is the input to a second level linear-kernel SVM. Our experiments on a dataset of 939 images categorized into 8 severity levels achieved 82% accuracy.