Short-time heavy rainfall is a kind of sudden strong and heavy precipitation weather, which seriously threatens people’s life and property safety. Accurate precipitation nowcasting is of great significance for the government to make disaster prevention and mitigation decisions in time. In order to make high-resolution forecasts of regional rainfall, this paper proposes a convolutional 3D GRU (Conv3D-GRU) model to predict the future rainfall intensity over a relatively short period of time from the machine learning perspective. Firstly, the spatial features of radar echo maps with different heights are extracted by 3D convolution, and then, the radar echo maps on time series are coded and decoded by using GRU. Finally, the trained model is used to predict the radar echo maps in the next 1-2 hours. The experimental results show that the algorithm can effectively extract the temporal and spatial features of radar echo maps, reduce the error between the predicted value and the real value of rainfall, and improve the accuracy of short-term rainfall prediction.

Ship target detection is an important guarantee for the safe passage of ships on the river. However, the ship image in the river is difficult to recognize due to the factors such as clouds, buildings on the bank, and small volume. In order to improve the accuracy of ship target detection and the robustness of the system, we improve YOLOv3 network and present a new method, called Ship-YOLOv3. Firstly, we preprocess the inputting image through guided filtering and gray enhancement. Secondly, we use k-means++ clustering on the dimensions of bounding boxes to get good priors for our model. Then, we change the YOLOv3 network structure by reducing part of convolution operation and adding the jump join mechanism to decrease feature redundancy. Finally, we load the weight of PASCAL VOC dataset into the model and train it on the ship dataset. The experiment shows that the proposed method can accelerate the convergence speed of the network, compared with the existing YOLO algorithm. On the premise of ensuring real-time performance, the precision of ship identification is improved by 12.5%, and the recall rate is increased by 11.5%.

As a generalization of several fuzzy tools, picture fuzzy sets (PFSs) hold a special ability to perfectly portray inherent uncertain and vague decision preferences. The intention of this paper is to present a Pearson’s picture fuzzy correlation-based model for multi-attribute decision-making (MADM) analysis. To this end, we develop a new correlation coefficient for picture fuzzy sets, based on which a Pearson’s picture fuzzy closeness index is introduced to simultaneously calculate the relative proximity to the positive ideal point and the relative distance from the negative ideal point. On the basis of the presented concepts, a Pearson’s correlation-based model is further presented to address picture fuzzy MADM problems. Finally, an illustrative example is provided to examine the usefulness and feasibility of the proposed methodology.

Massive sport, such as unmanned aerial vehicle performance, often needs fast and efficient calculation of formation morphing and individual path planning. This paper introduces a novel fast formation control method of a crowd. First, we get the agents’ location in a 2D polygon with centroidal Voronoi tessellation and L-BFGS techniques. Then, we transform crowd formation shapes with a global shortest motion path pair assignment using earth mover’s distance algorithm. Finally, the repulsing force between agents and obstacles is calculated based on the recursive velocity observer method control agents’ motion. Extensive experimental results show the effectiveness and usefulness of our algorithm in 2D group formation transformation.

Making unconventional emergent plan for dense crowd is one of the critical issues of evacuation simulations. In order to make the behavior of crowd more believable, we present a real-time evacuation route approach based on emotion and geodesic under the influence of individual emotion and multi-hazard circumstances. The proposed emotion model can reflect the dynamic process of individual in group on three factors: individual emotion, perilous field, and crowd emotion. Specifically, we first convert the evacuation scene to Delaunay triangulation representations. Then, we use the optimization-driven geodesic approach to calculate the best evacuation path with user-specified geometric constraints, such as crowd density, obstacle information, and perilous field. Finally, the Smooth Particle Hydrodynamics method is used for local avoidance of collisions with nearby agents in real-time simulation. Extensive experimental results show that our algorithm is efficient and well suited for real-time simulations of crowd evacuation.

The development of deep learning technology has promoted the wide application of face recognition in many scenarios such as mobile payment and social media, but the security of user data is facing great challenges. To protect the privacy of users, face authentication cannot be operated in plaintext. To solve this problem, a face feature ciphertext authentication scheme based on homomorphic encryption is proposed. First, the face image feature extraction is completed based on a deep learning model. Second, the face features are packaged into ciphertext by using homomorphic encryption and batch processing technology, and the face feature ciphertext is saved in the database of the cloud server. Third, combined with automorphism mapping and Hamming distance, a face feature ciphertext recognition method is designed, which can complete face recognition in the case of ciphertext. Finally, the integrity and consistency of face feature ciphertext recognition results before and after decryption are guaranteed by the one-time MAC authentication method. The whole framework can finish identity recognition without decrypting face feature coding, and the homomorphic ciphertext of face feature coding is saved in the database, so there is no risk of face feature coding leakage. Experiments show that the system has met the requirements of real application scenarios.

No abstract

High CT image quality is an important guarantee for doctors to correctly diagnose pulmonary nodules. The aim of this study was to explore the application value of PDCA management method in improving the quality of CT target scanning for pulmonary nodules. We identified 480 patients’ CT image with at least one pulmonary nodule admitted in Ninghai First hospital from September 1st, 2018, to April 30th, 2019. 240 CT images are carried out by the conventional target scanning method, and we analyzed the reasons for the low quality of some CT target scanning images of pulmonary nodules in the radiology department of our hospital. We established a new process of CT target scanning for pulmonary nodules based on the PDCA method and then tested 240 patients who were checked after January 1st, 2019. The excellent rate of CT target scanning image of pulmonary nodules in our department increased from 60.0% to more than 90.0%. The patients’ satisfaction with the examination was significantly higher than that without the implementation of PDCA management. The research result indicated that the process of CT target scanning image, postprocessing reconstruction, and numerical measurement of pulmonary nodules can be improved by standardized PDCA cycle, which benefits effectively improving the theoretical and operational skills of radiologists and significantly improving the image quality rate of CT target scanning of pulmonary nodules.