Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-12-13 Cooperative journals: 《计算机应用研究》
Abstract： In order to make better use of low-resolution face images, a new face super-resolution reconstruction based on L1/2 regularization and local texture constraints is proposed. In the process of face reconstruction, this paper uses Nonnegative Matrix Factorization to amplify the face image to an appropriate medium resolution. In this process, local texture constraints are used to enhance texture feature extraction. Then, face image reconstruction uses local sparse prior, and adds reconstruction constraints and local texture constraints again. In order to make the obtained face image sparse coefficient more sparse, L1/2 regularization is used to solve the problem. The experimental results show that the reconstructed face image retains the original image structure, and it can achieve good reconstruction results, and has better robustness.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-12-13 Cooperative journals: 《计算机应用研究》
Abstract： In order to solve the high transportation cost problem faced by the third party logistics (3PL) suppliers, from the fourth party logistics (4PL) supplier’s point of view, a multi 3PL Collaborative Routing Customization Problem is proposed. As the problem needs to consider the route and the 3PLs simultaneously, a hybrid Particle Swarm Optimization based K-shortest path algorithm (K-PSO) is designed. In the experimental analysis, the results show the effectiveness of the K-PSO by comparing with the GA and enumeration algorithm with different nodes’ and 3PL suppliers’ number. Finally, through the changes of the 3PL suppliers’ transit cost, it shows the advantages of the 4PL coordinated transportation.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-08-13 Cooperative journals: 《计算机应用研究》
Abstract： In order to solve the phenomenon of the edges of the image produce light bands in the de-hazing process and the whole image’s color becomes dim after de-hazing, this paper proposed a novel algorithm based on improved transmission and semi-inverse algorithm. Aiming at the dark channel prior failure at the edge of the image, a transmission correction mechanism is proposed in this paper to eliminate the band phenomenon at the edge and improve the application of the dark channel prior. Secondly, based on the atmospheric scattering model, the atmospheric light is obtained by using the semi-inverse algorithm. Then for the situation that the color of the whole image become dim after de-hazing, the image contrast algorithm based on wavelet transform is used to enhance the stretching treatment and improve the visual effect of image. The experimental results show that the algorithm can remove fog more effectively, and the ability to de-haze is better than the original algorithm.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-20 Cooperative journals: 《计算机应用研究》
Abstract： With the global aging population increasing, the daily behavior regulation and care of the old people have become a challenging social problem. In order to deal with the growing demand of society, this paper proposes a method that is driven by data and knowledge, and uses PSL (probabilistic soft logic) and multi-level analysis for modeling the old people's daily activities to solve the problem of nursing activity recognition. Experiments show that the method has higher accuracy than HMM in the event identification and abnormal activity detection, and has faster response speed than the non-hierarchical recognition method.
Hits
YES
Hits
Downloads
Comment