Search

Abstract With the arrival of new data acquisition platforms derived from the Internet of Things (IoT), this paper goes beyond the understanding of traditional remote sensing technologies. Deep fusion of remote sensing and computer vision has hit the industrial world and makes it possible to apply Artificial intelligence to solve problems such as automatic extraction of information and image interpretation. However, due to the complex architecture of IoT and the lack of a unified security protection mechanism, devices in remote sensing are vulnerable to privacy leaks when sharing data. It is necessary to design a security scheme suitable for computation‐limited devices in IoT, since traditional encryption methods are based on computational complexity. Visual Cryptography (VC) is a threshold scheme for images that can be decoded directly by the human visual system when superimposing encrypted images. The stacking‐to‐see feature and simple Boolean decryption operation make VC an ideal solution for privacy‐preserving recognition for large‐scale remote sensing images in IoT. In this study, the secure and efficient transmission of high‐resolution remote sensing images by meaningful VC is achieved. By diffusing the error between the encryption block and the original block to adjacent blocks, the degradation of quality in recovery images is mitigated. By fine‐tuning the pre‐trained model from large‐scale datasets, we improve the recognition performance of small encryption datasets for remote sensing images. The experimental results show that the proposed lightweight privacy‐preserving recognition framework maintains high recognition performance while enhancing security.

Abstract Aiming at the existing problems of metric-based methods, there are problems such as inadequate feature extraction, inaccurate class feature representation, and single similarity measurement. A new model based on attention mechanism and weight fusion strategy is proposed in this paper. Firstly, the image is passed through the conv4 network with channel attention mechanism and space attention mechanism to obtain the feature map of the image. On this basis, the fusion strategy is used to extract class-level feature representations according to the difference in contributions of different samples to class-level feature representations. Finally, the similarity scores of query set samples are calculated through the network to predict the classification. Experimental results on the miniImageNet dataset and the omniglot dataset demonstrate the effectiveness of the proposed method.

Remote sensing images (RSIs) contain important information, such as airports, ports, and ships. By extracting RSI features and learning the mapping relationship between image features and text semantic features, the interpretation and description of RSI content can be realized, which has a wide range of application value in military and civil fields, such as national defense security, land monitoring, urban planning, and disaster mitigation. Aiming at the complex background of RSIs and the lack of interpretability of existing target detection models, and the problems in feature extraction between different network structures, different layers, and the accuracy of target classification, we propose an object detection and interpretation model based on gradient-weighted class activation mapping and reinforcement learning. First, ResNet is used as the main backbone network to extract the features of RSIs and generate feature graphs. Then, we add the global average pooling layer to obtain the corresponding feature weight vector of the feature graph. The weighted vectors are superimposed to output class activation maps. The reinforcement learning method is used to optimize the generated region generation network. At the same time, we improve the reward function of reinforcement learning to improve the effectiveness of the region generation network. Finally, network dissecting analysis is used to obtain the interpretable semantic concept in the model. Through experiments, the average accuracy is more than 85%. Experimental results in the public RSI description dataset show that the proposed method has high detection accuracy and good description performance for RSIs in complex environments.

Aiming at facial expression recognition, the recognition rate is not high due to noise and occlusion. A hybrid approach of facial expression has been presented by combining local and global features. First, feature extraction is performed to fuse the histogram of oriented gradients (HOG) descriptor with the compounded local ternary pattern (C-LTP) descriptor. Second, features extracted by HOG and C-LTP are fused into a single feature vector. Third, the feature vector is sent to a multi-class support vector machine classifier for facial classification. Finally, the proposed method is compared with the existing facial expression recognition methods in three public facial expression image databases, and the results show that the recognition rates of the proposed method in MMI, JAFFE and CK[KG-*2]+ databases are 98.28%, 95.75% and 99.64%, respectively. The average recognition rate is 10% higher than other methods, which is better than other existing methods. The results of this study provide a reference for the research of facial expression recognition in many situations. The method of facial expression recognition proposed can effectively promote the development of human-computer interaction system and the study of computer image understanding. It is of great significance to realize the fusion of human language and natural language, as well as the establishment and implementation of the connection model between language and expression.

Network intrusion can enter the network through informal channels. Some illegal users utilize Trojans and self-programmed attack to change the network security system, so that the system loses the defense and alarm function and the Hacker can steal the internal information. Network intrusion seriously harms the security of network information and the legitimate rights of users. Therefore, a bagging strategy-based kernel extreme learning machine for complex network intrusion detection is presented in this paper. This method adopts a bagging strategy to train several sub-kernel extreme learning machines independently. Then the integrated gain of above machines is measured based on the margin distance minimization (MDM) criterion. Selected machines with high gain degree are selected for selective integration to obtain selective integrated learners with strong generalization ability and high efficiency. Then an improved universal gravitation search algorithm is used to optimize the kernel parameters. Meanwhile, a sub-kernel extreme learning machine online update strategy based on incremental learning of batch samples is introduced to realize the online update of intrusion detection model, so that the proposed detection method can effectively be adapted to the changes of complex network environment. Finally, experiments illustrate that the proposed method has better effect on network intrusion detection in terms of detection accuracy and speed, especially for unknown network intrusion connection events, the response speed is fast, the false alarm rate is low.

With the rapid development of social network, intelligent terminal and automatic positioning technology, location-based social network (LBSN) service has become an important and valuable application. Point of interest (POI) recommendation is an important content in LBSN, which aims to recommend new locations of interest for users. It can not only alleviate the information overload problem faced by users in the era of big data, improve user experience, but also help merchants quickly find target users and achieve accurate marketing. Most of the works are based on users' check-in history and social network data to model users' personalized preferences for interest points, and recommend interest points through collaborative filtering and other recommendation technologies. However, in the check-in history, the multi-source heterogeneous information (including the position, category, popularity, social, reviews) describes user activity from different aspects which hides people's life style and personal preference. However, the above methods do not fully consider these factors' combined action. Considering the data privacy, it is difficult for individuals to share data with others with similar preferences. In this paper, we propose a privacy protection point of interest recommendation algorithm based on multi-exploring locality sensitive hashing (LSH). This algorithm studies the POI recommendation problem under distributed system. This paper introduces a multi-exploring method to improve the LSH algorithm. On the one hand, it reduces the number of hash tables to decrease the memory overhead; On the other hand, the retrieval range on each hash table is increased to reduce the time retrieval overhead. Meanwhile, the retrieval quality is similar to the original algorithm. The proposed method uses modified LSH and homomorphic encryption technology to assist POI recommendation which can ensure the accuracy, privacy and efficiency of the recommendation algorithm, and it verifies feasibility through experiments on real data sets. In terms of root mean square error (RMSE), mean absolute error (MAE) and running time, the proposed method has a competitive advantage.

As the continuous development of mobile social networks, the structure of the mobile social network increasingly becomes complex. It not only speeds up information transmission between people but also expands the scope of information exchange, which has become an essential and important social media in people's social life. How to effectively identify and classify these online communities has important practical significance for the study of social networks. Correctly detecting the community structure of mobile social networks can not only improve the accuracy of friend recommendation, link prediction, service user positioning, product marketing, and other aspects but also provide an important basis for the monitoring of online public opinion. But the traditional social network cluster method based on the trust degree mainly calculates the user trust by analyzing the interactive feedback information between users. This method cannot effectively solve the “cold start” problem in the trust calculation process, that is, for the new network node, the trust value of this node cannot be accurately measured due to the lack of interaction with other nodes. Focusing on this problem, we propose a Gaussian pigeon-oriented graph clustering algorithm for social networks' cluster in this paper. A graph model is first built. Then, an efficient K-medoid algorithm is utilized to search the user center in all groups. The Gaussian pigeon algorithm is used to search the similarity between each user and the central user. Users that meet the similarity threshold are divided into the same user group. Finally, the simulation results show that the proposed method has better cluster effect than other state-of-the-art social networks' clustering approaches.

With the ever-increasing amount and complexity of remote sensing image data, the development of large-scale image segmentation analysis algorithms has not kept pace with the need for methods that improve the final accuracy of object recognition. In particular, the development of such methods for large-scale images poses a considerable challenge. Traditional image segmentation methods are of high time complexity and low segmentation accuracy. This paper presents a new large-scale remote sensing image segmentation method that combines fuzzy region competition and the Gaussian mixture model. The new algorithm defines a non-similarity measurement for the class attribute of pixels using the Gaussian mixture model. This algorithm has the ability to perform high precision fitting of data with a statistical distribution, thereby effectively eliminating the influence of noise on the segmentation result. Moreover, fuzzy region competition is introduced to define the prior probability of the neighborhood that will be regarded as the weight of the Gaussian component. This process enhances the robustness of large-scale image segmentation. Finally, we acquire the image data from Google Earth and conduct experiments; the experimental results show that this new method has the feasibility and effectiveness and can achieve highly accurate segmentation results compared with current state-of-the-art methods.

Currently, big data is a new and hot issue. Particularly, the rapid growth of the Internet of Things causes a sharp growth of data. Enormous amounts of networking sensors are continuously collecting and transmitting data to be stored and processed in the cloud, including remote sensing data, environmental data, and geographical data. And region is regarded as the very important object in remote sensing data, which is mainly researched in this article. Region search is a crucial task in remote sensing process, especially for military area and civilian fields. It is difficult to fast search region accurately and achieve generalizability of the regions’ features due to the complex background information, as well as the smaller size. Especially, when processing region search in large-scale remote sensing image, detailed information as the feature can be extracted in inner region. To overcome the above difficulty region search task, we propose an accurate and fast region search in optical remote sensing images under cloud computing environment, which is based on hybrid convolutional neural network. The proposed region search method partitioned into four processes. First, fully convolutional network is adopted to produce all the candidate regions that contain the possible object regions. This process avoids exhaustive search for input images. Then, the features of all candidate regions are extracted by a fast region-based convolutional neural network structure. Third, we design a new difficult sample mining method for the training process. At the end, in order to improve the region search precision, we use an iterative bounding box regression algorithm to normalize the detected bounding boxes, in which the regions contain candidate objects. The proposed algorithm is evaluated on optical remote sensing images acquired from Google Earth. Finally, we conduct the experiments, and the obtained results show that the proposed region search method constantly achieves better results regardless of the type of images tested. Compared with traditional region search methods, such as region-based convolutional neural network and newest feature extraction frameworks, our proposed methods show better robustness with complex context semantic information and backgrounds.

Remote sensing image object detection is widely used in civil and military fields. The important task is to detect objects such as ships, planes, airports, harbours and so on, and then it can obtain object category and position information. It is of great significance to use remote sensing images to observe the densely arranged and directional targets such as cars and ships parked in parking lots and harbours. The object detection task mainly includes object localization and classification. Remote sensing images contain large number of small objects and dense scenes due to the long shooting distance and wide coverage. Small objects occupy few pixels in the image, and they are easily miss‐detected. In dense scenes, the overlapping part of each object is large, so it is easy to detect objects repeatedly. The traditional small object detection methods deliver low accuracy and take long time. Therefore, object detection is very challenging. We put forward a novel deep learning‐based single shot multibox detector (SSD) model for object detection. First, we propose an improved inception network to optimize SSD to strengthen the small object feature extraction ability (FEA) in the shallow network. Second, the feature pyramid network is modified to enhance the fusion effect. Third, the deep feature fusion module is designed to improve the FEA of the deep network. Finally, the extracted image features are matched with candidate boxes with different aspect ratios to perform object detection and location with different scales. Experiments on DOTA show that the proposed algorithm can adapt to the remote sensing object detection in different backgrounds, and effectively improve the detection effect of remote sensing objects in complex scenes. [ABSTRACT FROM AUTHOR]