Aiming at the problems of low accuracy and poor real-time performance of Noctiluca scintillans red tide extraction in the field of satellite remote sensing, a Noctiluca scintillans red tide extraction method from Unmanned Aerial Vehicle (UAV) images based on deep learning was proposed. Firstly, the high-resolution RGB (Red-Green-Blue) videos collected by UAV were used as the monitoring data, the backbone network was modified to VGG-16 (Visual Geometry Group-16) and the spatial dropout strategy was introduced on the basis of the original UNet++ network to enhance the feature extraction ability and prevent the overfitting respectively. Then, the VGG-16 network pre-trained by using ImageNet dataset was applied to perform transfer learning to increase the network convergence speed. Finally, in order to evaluate the performance of the proposed method, experiments were conducted on the self-built red tide dataset Redtide-DB. The Overall Accuracy (OA), F1 score, and Kappa of the Noctiluca scintillans red tide extraction of the proposed method are up to 94.63%, 0.955 2, 0.949 6 respectively, which are better than those of three traditional machine learning methods — K-Nearest Neighbors (KNN), Support Vector Machine (SVM) and Random Forest (RF) as well as three typical semantic segmentation networks (PSPNet (Pyramid Scene Parsing Network), SegNet and U-Net). Meanwhile, the red tide images of different shooting equipment and shooting environments were used to test the generalization ability of the proposed method, and the corresponding OA, F1 score and Kappa are 97.41%, 0.965 9 and 0.938 2, respectively, proving that the proposed method has a certain generalization ability. Experimental results show that the proposed method can realize the automatic accurate Noctiluca scintillans red tide extraction in complex environments, and provides a reference for Noctiluca scintillans red tide monitoring and research work.
In the era of big data, research in topic evolution is mostly based on the classical probability topic model, the premise of word bag hypothesis leads to the lack of semantic in topic and the retrospective process in analyzing evolution. An online incremental feature ontology based topic evolution algorithm was proposed to tackle these problems. First of all, feature ontology was built based on word co-occurrence and general WordNet ontology base, with which the topic in text stream was modeled. Secondly, a text stream topic matrix construction algorithm was put forward to realize online incremental topic evolution analysis. Finally, a text topic ontology evolution diagram construction algorithm was put forward based on the text steam topic matrix, and topic similarity was computed using sub-graph similarity calculation, thus the evolution of topics in text stream was obtained with time scale. Experiments on scientific literature showed that the proposed algorithm reduced time complexity to O(nK+N), which outperformed classical probability topic evolution model, and performed no worse than sliding-window based Latent Dirichlet Allocation (LDA). With ontology introduced, as well as the semantic relations, the proposed algorithm can demonstrate the semantic feature of topics in graphics, based on which the topic evolution diagram is built incrementally, thus has more advantages in semantic explanatory and topic visualization.