• Volume 31,Issue 5,2022 Table of Contents
    Select All
    Display Type: |
    • >Survey
    • Overview on Graph-based Zero-shot Learning

      2022, 31(5):1-20. DOI: 10.15888/j.cnki.csa.008463

      Abstract (3644) HTML (1724) PDF 2.46 M (2223) Comment (0) Favorites

      Abstract:Although the deep learning method has made a huge breakthrough in machine learning, it requires a large amount of manual work for data annotation. Limited by labor costs, however, many applications are expected to reason and judge the instance labels that have never been encountered before. For this reason, zero-shot learning (ZSL) came into being. As a natural data structure that represents the connection between things, the graph is currently drawing more and more attention in ZSL. Therefore, this study reviews the methods of graph-based ZSL systematically. Firstly, the definitions of ZSL and graph learning are outlined, and the ideas of existing solutions for ZSL are summarized. Secondly, the current ZSL methods are classified according to different utilization ways of graphs. Thirdly, the evaluation criteria and datasets concerning graph-based ZSL are discussed. Finally, this study also specifies the problems to be solved in further research on graph-based ZSL and predicts the possible directions of its future development.

    • Survey on Multikernel Operating Systems

      2022, 31(5):21-29. DOI: 10.15888/j.cnki.csa.008426

      Abstract (807) HTML (1327) PDF 1.27 M (2476) Comment (0) Favorites

      Abstract:Operating systems play an important role in modern life. To serve different hardware environments and diverse application scenarios, they need to be scalable and flexible while maintaining good performance. A multicore operating system, as a kind of distributed operating system, is one of the solutions to this problem. This study analyzes the design principles of multicore operating systems, investigates existing multicore operating system technologies, and compares these technologies with other relevant ones. Finally, the current situation and trend of the research on multicore operating systems are summarized.

    • Leaf Vein Segmentation Method of Soybean Leaf Images

      2022, 31(5):30-39. DOI: 10.15888/j.cnki.csa.008467

      Abstract (600) HTML (1093) PDF 2.44 M (1251) Comment (0) Favorites

      Abstract:Leaf vein segmentation is an important step in leaf pattern analysis, which is of great significance for soybean variety identification and phenotype research. On account of the complicated vein structure of soybean leaves and the low contrast of the leaf area where the veins are located, it is generally impossible to achieve ideal leaf vein segmentation results only using gray information. This study presents a soybean vein segmentation method combining the multi-scale gray unconstrained hit-or-miss transform (UHMT) algorithm and the processing method based on the hue data of HSI color space. In this method, the gray information in RGB color space and the hue data in HSI color space are used to segment the global leaf veins and local primary and secondary veins from soybean leaf images, respectively. The former uses iterative threshold segmentation to extract the leaf area and eliminates interference factors such as the outer contour and the petiole through expansion and corrosion to obtain the leaf area image. Then, the multi-scale gray UHMT algorithm is employed to obtain the global leaf vein image. Considering the poor performance of primary and secondary vein segmentation, we use hue data to enlarge the discrepancies in gray values between veins pixels and other pixels to realize the segmentation of local primary and secondary veins. The obtained global and local vein images are fused into the final soybean leaf vein image. Moreover, this study utilizes soybean leaf images in the soybean leaf image database, SoyCultivar, to verify the effectiveness of the algorithm. The results indicate that this algorithm is better than existing leaf vein segmentation methods as it can not only extract soybean leaf veins completely but also well eliminate the background, leaf contours, petioles, and other irrelevant components.

    • >Survey
    • Survey on Single Object Tracking Algorithms Based on Deep Learning

      2022, 31(5):40-51. DOI: 10.15888/j.cnki.csa.008476

      Abstract (915) HTML (1739) PDF 2.41 M (2051) Comment (0) Favorites

      Abstract:Single object tracking is a research focus in the field of computer vision. Traditional algorithms including correlation filtering have fast tracking speed but generally low tracking accuracy due to the roughness of extracted manual features such as color and gray levels. With the development of deep learning theory in recent years, tracking methods using deep features can achieve a good balance between tracking accuracy and speed. This study first introduces the relevant background of single object tracking and then sorts out multiple algorithms that have emerged in the development of single object tracking from the two stages of single object tracking based on correlation filters and deep learning. The current mainstream Siamese network algorithms are also introduced in detail. Finally, a large data set is used to compare and analyze the excellent algorithms that have emerged in recent years. In view of the shortcomings and deficiencies of these algorithms, the development prospects of this field are provided in this study.

    • Improved Deep Neural Network with Overlapped Time Slice for Pattern Recognition of Motor Imagery EEG

      2022, 31(5):52-64. DOI: 10.15888/j.cnki.csa.008459

      Abstract (424) HTML (1517) PDF 2.67 M (931) Comment (0) Favorites

      Abstract:Pattern recognition of electroencephalogram (EEG) signals during motor imagery (MI) has been one of the most important research directions in the field of non-invasive brain-computer interface (BCI). In recent years, deep learning has further improved the recognition accuracy of EEG signals during MI. However, given the strong time variability of EEG signals, there are still some problems such as insufficient training samples and too high feature dimensions. To solve the above problems, this study proposes a new training strategy called “overlapped time slice”. Based on the existing cropped time slice strategy, this study adopts a novel overlapped time slice strategy and constructs a new loss function calculation and label prediction method with the overlapped time slice set. The overlapped time slice strategy can not only further increase the number of training samples but also reduce the feature space of a single sample to improve the performance of the deep neural network in EEG signal recognition. For the verification of the feasibility and effectiveness of the proposed overlapped strategy, three open-source EEG signal datasets, namely the BCI Competition IV datasets 1, 2a, and 2b, are selected in this study, and five kinds of deep neural network models are built on these three datasets. During experiments, the performance and efficiency of MI recognition are compared between the cropped strategy and the overlapped strategy. Experimental results show that the overlapped strategy has better recognition performance than that of the cropped strategy. Finally, nine groups of experiments are designed with different parameter combinations by adjusting the parameters of the overlapped time slice strategy. The experimental results demonstrate that parameter combination affects the final classification performance and that the classification performance is not in a linear relationship with the efficiency. The recognition accuracy of the proposed overlapped strategy on dataset 1, 2a, and 2b is 92.3%, 77.8%, and 86.3% respectively. Compared with the conventional cropped strategy, the proposed overlapped strategy has improved the performance significantly without necessarily reducing the efficiency.

    • Technology of Quantum-resistant Trusted Computing Security Support Platform

      2022, 31(5):65-74. DOI: 10.15888/j.cnki.csa.008550

      Abstract (585) HTML (904) PDF 1.32 M (1074) Comment (0) Favorites

      Abstract:With the development of science and technology, the deployment of large-scale quantum computers is becoming possible, and the public-key cryptographic algorithms based on some difficult problems will be solved by quantum algorithms effectively. The security of traditional trusted hardware chips such as TCM/TPM will be seriously affected due to the wide use of public-key cryptosystems such as RSA, SM3, and ECC, and most of the quantum-resistant (QR) cryptographic algorithms cannot be implemented on hardware chips with limited computational resources. Therefore, it is necessary to redesign the QR trusted computing platform. In this study, considering the security challenges faced by trusted computing in quantum computing models, we summarize the present situation of QR trusted computing research and propose a QR trusted computing technology system. Combined with the existing post-quantum cryptographic protocol and trusted computing software and hardware technology framework, we transplant the QR cryptographic algorithms and protocol on the trusted computing platform and implement a prototype system of a QR trusted computing security support platform based on TCM. The work includes the design of the primitive root key and QR extensions such as TCM cipher library, remote attestation, and LDAA. Finally, the results of function and performance tests on the emulator for the above TCM modules show that the prototype system is resistant to attacks by quantum algorithms, with acceptable application performance overhead.

    • Design of Combustion Dynamics Platform

      2022, 31(5):75-84. DOI: 10.15888/j.cnki.csa.008464

      Abstract (501) HTML (650) PDF 3.52 M (864) Comment (0) Favorites

      Abstract:The research on fuel plays an important role in engine development. Data is the basis of combustion model coupling, and the reaction mechanism is the key to combustion numerical simulation. This study designs a combustion dynamics platform including a combustion database and an online calculation system of combustion reaction mechanisms to address the shortcomings in the construction of combustion databases in China and satisfy the urgent demand for the construction of personalized combustion reaction mechanisms. The combustion database system establishes an efficient search process through data standardization and hierarchical management and forms a unique multi-type theoretical and experimental data retrieval system in China. High-quality data is absorbed into the platform through data collection and evaluation. The online calculation system of combustion reaction mechanisms establishes a one-stop calculation process of software input, file creation, calculation, and result analysis through the Web-based interactive molecular modeling. The construction and application of the combustion dynamics platform will reduce the dependence on foreign data platforms, promote the exchange and sharing of domestic combustion reaction data, significantly lower the threshold for the use of fuel reaction mechanism software, and further deepen the innovative research.

    • IoT-oriented Edge Node Platform Architecture

      2022, 31(5):85-93. DOI: 10.15888/j.cnki.csa.008532

      Abstract (781) HTML (762) PDF 1.82 M (794) Comment (0) Favorites

      Abstract:The structure of the edge node system is exposed to problems of the single function, the confusing module division, and the high bandwidth overhead, and security risks in edge-to-cloud data transmission. To solve these problems, this study designs and develops an IoT-oriented edge node platform architecture. The platform architecture consists of five modules, i.e., device management, data management, resource management, application management, and exporting service management. The designed platform architecture applies algorithms such as the least square method to achieve functions including data collection, data filtering, data encryption, and indicator analysis on the edge side. The experimental results indicate that the platform can continuously collect surroundings information, filter and analyze data, and ensure the safe and reliable transmission of data while reducing the cost of edge-to-cloud data transmission.

    • Design of Monitoring Platform of Heavy-duty Vehicle Pollutant Emission Based on Distributed Micro-service Architecture

      2022, 31(5):94-101. DOI: 10.15888/j.cnki.csa.008451

      Abstract (412) HTML (448) PDF 1.93 M (728) Comment (0) Favorites

      Abstract:As the problem of vehicle exhaust pollution becomes increasingly serious, the Ministry of Ecology and Environment of the People’s Republic of China requires all China VI heavy-duty vehicles to install onboard remote emission management terminals to monitor their exhaust emissions. This study designs a zoning monitoring platform of heavy-duty vehicle pollutant emissions based on a distributed micro-service architecture. The platform collects real-time data of all the heavy-duty vehicles with onboard terminals in a designated area and quantifies the main exhaust pollutants of heavy-duty vehicles, namely nitrogen oxides and particulates. Meanwhile, an algorithm determining the specific administrative district to which a pollutant belongs is proposed to count and display the total pollutant emissions of heavy-duty vehicles in each administrative district. The platform is deployed in the Zibo Environmental Protection Bureau. Operating normally with reliable data, it has provided strong data support for the precise monitoring and governance of environmental protection departments.

    • Research and Application of Ancient Poetry Learning System on Mobile Terminal

      2022, 31(5):102-110. DOI: 10.15888/j.cnki.csa.008504

      Abstract (545) HTML (1461) PDF 2.70 M (961) Comment (0) Favorites

      Abstract:With the development of the mobile Internet, the mobile knowledge acquisition mode has become the new favorite of the times. Poetry is a bright pearl of Chinese culture. Therefore, the combination of poetry learning and mobile learning is urgent. The proposed system adopts the architecture of Client/Server and uses the Faster R-CNN model to achieve image identification. Then, the function of generating ancient poems is performed by the recurrent neural network (RNN) model, and personalized recommendation is implemented through the collaborative filtering recommendation algorithm. The client APP is developed on the basis of the Flutter and SpringBoot frameworks. The database is managed by the relational database management system MySQL and connected to the system through the server to fulfill the desired functions. For those who have needs or strong interest in learning poetry, a poetry learning system committed to carrying forward the Chinese culture is developed by leveraging image recognition and deep learning technologies to achieve intelligent image recognition and generation of ancient poetry.

    • Environmental Monitoring System Based on Microservice Architecture and Blockchain

      2022, 31(5):111-117. DOI: 10.15888/j.cnki.csa.008473

      Abstract (427) HTML (630) PDF 1.13 M (902) Comment (0) Favorites

      Abstract:In response to the main challenges and problems facing the environmental monitoring of expressway service areas, this study proposes a method of constructing an environmental monitoring system based on the microservice architecture combined with Blockchain technology and develops an intelligent environmental monitoring and management system for expressways. Through the Internet of Things (IoT) middleware technology, the barriers for the transmission of environmental monitoring data from terminal sensors to cloud servers are broken. The overall system is divided into water quality, air, noise, and solid waste service units (can be deployed independently) in light of the divide-and-conquer method. The storage management and retrieval of massive structured data are realized using multiple data sources combined with time-stamp horizontal fragmentation. The reliable storage and verification of monitoring and alarm data are completed with the help of blockchain technology. Finally, the visual display is enabled by mobile Internet technology.

    • Application of Deep Learning in Mixed Reality Workshop Inspection

      2022, 31(5):118-123. DOI: 10.15888/j.cnki.csa.008515

      Abstract (395) HTML (586) PDF 1.12 M (778) Comment (0) Favorites

      Abstract:The traditional method of workshop inspection mainly relies on manual check and recording, which is cumbersome and cannot be shared in real time. For higher work efficiency, deep learning is applied to mixed reality workshop inspection. It is combined with mixed reality technology, and the ResNet network is used to classify and identify workshop equipment. After classification and identification, HoloLens’ spatial perception ability is leveraged to locate and confirm the equipment. Finally, equipment basic information, operating status, and alarms are displayed. The experimental results show that compared with traditional workshop inspection methods, ResNet, with a high identification rate, can effectively filter noises, improve the utilization rate and identification rate of HoloLens, and consequently improve the work efficiency of inspection personnel.

    • Bidirectional Identity Verification Scheme of Cross-regional Medical Collaboration Based on SM9 Algorithm

      2022, 31(5):124-130. DOI: 10.15888/j.cnki.csa.008503

      Abstract (400) HTML (445) PDF 1.51 M (828) Comment (0) Favorites

      Abstract:For the safety of cross-regional medical collaborative services, it is crucial to identify the identities of the exchange parties. This study combines the Diffie-Hellman (DH) algorithm with the SM9 algorithm published by the State Cryptography Administration of China, uses the common key negotiated by the DH algorithm as a verification factor, and integrates digital signatures to achieve encrypted transmission and bidirectional verification of users. Taking the identity verification of users in the access of electronic medical records between hospitals as a case, this study analyzes the of both communicating parties in cross-regional information sharing and verifies the feasibility and correctness of the scheme through experiments, demonstrating that the scheme has practical application value.

    • Snapshot System for Linux Non-logical Volume Block Devices

      2022, 31(5):131-136. DOI: 10.15888/j.cnki.csa.008452

      Abstract (368) HTML (562) PDF 1.07 M (788) Comment (0) Favorites

      Abstract:Given that non-logical volume block devices under the Linux operating system need to create temporary snapshots without additional block devices being added to store data, this study designs and implements a snapshot system for Linux non-logical volume block devices. The system has an application layer and a generic block layer of the inner nuclear layer and is based on copy-on-write (COW). The application layer analyzes the user’s creation or deletion commands and transmits them to the generic block layer. The general block layer creates or deletes the snapshot devices and intercepts the general block layer I/O (bio) request of the source device and performs COW after the snapshot is created. Experimental results show that the system can create snapshots correctly. The optimal copy block size is 4 MB. The minimum impact on the write performance of the source device is less than 10%.

    • Logo Detection Method Combining Coordinate Attention and Adaptive Residual Connection

      2022, 31(5):137-146. DOI: 10.15888/j.cnki.csa.008462

      Abstract (354) HTML (655) PDF 1.58 M (675) Comment (0) Favorites

      Abstract:Logo detection has a wide range of applications in areas such as brand recognition and intellectual property protection. In order to solve problems of poor detection performance on small-scale logo and inaccurate logo positioning, a logo detection method is proposed based on the YOLOv4 network. Five continuous convolutional layers in the PANet module of YOLOv4 network are replaced by the designed adaptive residual blocks to enhance the utilization of shallow and deep features and fuse features with emphasis and optimize the model training. And the coordinate attention mechanism is used after the adaptive residual blocks to encode channel relationship and long-term dependencies through precise location information, filter and enhance the more useful features from the fused features. The K-means++ clustering algorithm is used to obtain anchor boxes which are more suitable for the logo datasets and assign those to different feature scales. The experimental results show that the mean average precision of the proposed method on FlickrLogos-32 and FlickrSportLogos-10 datasets reaches 88.09% and 84.72%, which is 0.91% and 1.40% higher than the original algorithm, respectively. The performance of the proposed method in positioning accuracy and small-scale logo detection is significantly improved.

    • Sentiment Polarity Analysis of Customer Dialogues Based on Graph Convolutional Network

      2022, 31(5):147-156. DOI: 10.15888/j.cnki.csa.008524

      Abstract (396) HTML (498) PDF 1.97 M (818) Comment (0) Favorites

      Abstract:With the development of power business, a large amount of data is produced in the link of customer service. However, traditional sentiment analysis methods for dialogues face many problems and challenges in customer service quality detection. In this study, the word graph is constructed according to the arrangement and location of words, and then the discontinuous long-distance semantic modeling of the whole sentence is carried out. Next, according to the relationship among different parts of the document, the self and interaction dependency relationships between sentence contexts are modeled, respectively. Finally, the convolutional neural network (CNN) is applied to the constructed graph for feature extraction and feature aggregation of the neighbor nodes to obtain the final feature representation of the text. In this way, the detection of emotional states is realized in customer dialogues. Experimental results show that the performance of the proposed model is always higher than that of the baseline model, which demonstrates that the fusion of word co-occurrence relationships, as well as sequential context coding and interactive context coding structures, can effectively improve the accuracy of sentiment category detection. This method provides a fine-grained analysis for intelligently and automatically detecting the emotional states in customer dialogues, which is of great significance to effectively improve the quality of customer service.

    • Improved Algorithm for Constructing K-degree Anonymous Graphs

      2022, 31(5):157-164. DOI: 10.15888/j.cnki.csa.008466

      Abstract (364) HTML (449) PDF 1.64 M (867) Comment (0) Favorites

      Abstract:In social networks, anonymization processing is required to prevent the leakage of user privacy before user data is released. In this study, the social network anonymization is modeled as the k-degree anonymization of graphs given the privacy attack with the background of node degrees. The major method is to add as few edges or nodes as possible to the graph to meet the degree-anonymization requirements, and by doing this, the structural characteristics of the original graph are expected to be maintained to a large extent. At present, the edge-adding algorithms cannot well retain the basic structural characteristics such as the average path length, while the edge-node adding algorithms can better retain the structural characteristics of the original graph after adding many edges or nodes. Considering this situation, this study proposes an improved algorithm combining the two algorithms. Firstly, the greedy algorithm is used to generate the anonymity sequence, and then the edge-adding operation is performed on the basis of the community structure. The anonymization requirements of nodes with higher anonymity costs than the average anonymity costs are satisfied first, and node adding can be applied to the graph when edge adding cannot complete the anonymization. The experimental results of actual datasets show that the new algorithm can better retain several typical structural characteristics of the graph, and the number of added edges or nodes is less.

    • Recognition of Vehicle Abnormal Driving Behaviors Based on LSTM-att

      2022, 31(5):165-173. DOI: 10.15888/j.cnki.csa.008478

      Abstract (456) HTML (1917) PDF 2.76 M (889) Comment (0) Favorites

      Abstract:Abnormal behaviors of vehicles may cause traffic accidents or even economic losses and casualties. Accurate recognition of abnormal vehicle behaviors can prevent potential hazards. To tackle the problems in existing studies, such as difficulty to retain the time characteristics of data, this study proposes a recognition model of long short-term memory (LSTM) neural network with an attention layer, and trains and verifies the proposed model by using abnormal vehicle trajectories in real traffic scenes. The experimental results show that the proposed model can effectively recognize abnormal driving behaviors with accuracy reaching 98.4%.

    • Low-light Image Enhancement Algorithm Based on GAN and U-Net

      2022, 31(5):174-183. DOI: 10.15888/j.cnki.csa.008431

      Abstract (903) HTML (2613) PDF 2.05 M (1396) Comment (0) Favorites

      Abstract:The image data generated at night, under low light conditions, etc., has the problems of too dark images and loss of details, which hinders the understanding of image content and the extraction of image features. The research of enhancing this type of images to restore the brightness, contrast and details is meaningful in the applications of digital photography and upstream computer vision tasks. This study proposes a U-Net-based generative adversarial network. The generator is a U-Net model with a hybrid attention mechanism. The hybrid attention module can combine the asymmetric Non-local global information and the channel weight information of channel attention to improve the feature representation ability of the network. A fully convolutional network model based on PatchGAN is taken as the discriminator to perform local processing on different regions of the images. We introduce a multi-loss weighted fusion method to guide the network to learn the mapping from low-light images to normal-light images from multiple angles. Experiments show that this method achieves better results regarding objective indicators such as peak signal-to-noise ratio and structural similarity and reasonably restores the brightness, contrast and details of the images to intuitively improve their perceived quality.

    • 3D Path Planning of UAV in Complex Urban Environment

      2022, 31(5):184-194. DOI: 10.15888/j.cnki.csa.008514

      Abstract (687) HTML (1741) PDF 2.34 M (1105) Comment (0) Favorites

      Abstract:A transition-based rapidly-exploring random tree (T-RRT) algorithm can quickly find a low-risk path for a robot in a two-dimensional complex cost space, but it delivers a poor planning result for an unmanned aerial vehicle (UAV) in the three-dimensional flight condition. To solve this problem, this study proposes an exploring heuristic transition-based RRT (EHT-RRT) algorithm. The algorithm introduces the heuristic idea of the A* algorithm on the basis of the T-RRT to explore the heuristic cost, and it estimates the path cost from the perspectives of risk degree, path length, path deflection angle, and height change to improve the quality of the path. Then, the local node slip strategy is employed to make the path deviate to the low-risk area, and the local best direction attribute is added to each node. At last, the tree node exploration mechanism is improved through three directional vectors, i.e., random direction, target direction, and local best direction, to get rid of the blindness of the T-RRT algorithm in path finding. In addition, a target point offset with a probability of 20% is used to improve the planning efficiency. The results of simulation experiments show that compared with T-RRT, BT-RRT, and T-RRT* algorithms with the same target point offset each, the EHT-RRT algorithm can generate a shorter, safer, and smoother 3D path and better solve the 3D path planning problem of UAV in complex urban environments.

    • Sentiment Analysis Method Based on Temporal Multimodal Data Between Utterances

      2022, 31(5):195-202. DOI: 10.15888/j.cnki.csa.008475

      Abstract (332) HTML (794) PDF 1.20 M (817) Comment (0) Favorites

      Abstract:The traditional sentiment analysis methods based on single-modal data have always had problems such as a single analysis angle and low classification accuracy. The analysis method based on temporal multimodal data provides the possibility to solve these problems. On the basis of the temporal multimodal data between utterances, this study improves the existing multimodal sentiment analysis method and uses the bidirectional gated recurrent unit (Bi-GRU) combined with the intra-modal and cross-modal context attention mechanism for sentiment analysis. The sentiment analysis is finally verified on the MOSI and MOSEI datasets. Experiments show that the method of using temporal multimodal data between utterances and fully integrating intra-modal and cross-modal context information can be applied to sentiment analysis from the perspective of multimodal and temporal features. By doing this, the classification accuracy of sentiment analysis can be effectively improved.

    • Prediction of Runway Visual Range Based on LSTM

      2022, 31(5):203-212. DOI: 10.15888/j.cnki.csa.008492

      Abstract (372) HTML (498) PDF 1.97 M (730) Comment (0) Favorites

      Abstract:Runway visual range (RVR) reflects the pilot’s visual range, which is one of the important meteorological elements to ensure aircraft flight safety when the aircrafts take off and land. Improving the prediction accuracy of RVR will effectively improve the aircraft’s take-off and landing ability and aviation control ability under low visibility and complex weather conditions. The RVR is mainly affected by fog, smoke, dust, heavy precipitation and other weather, as well as the lack of instruments. According to the time series data of meteorological elements such as wind speed, humidity, temperature and runway visibility observed by the civil aviation automated weather observing system of Xianyang Airport from 2012 to 2018, this study firstly analyzes the long-term correlation relationship between the RVR and other meteorological observation data. On the basis of the correlation analysis, this study also uses the long short term memory network (LSTM), which is the most commonly used in artificial intelligence field, to construct an airport RVR prediction model. The experiment results show that the average fitting degree of the model can reach 72% within 0–2 h.

    • Segmentation and Quantification Method of Machine-made Sand Powder Based on Improved UNet Network

      2022, 31(5):213-221. DOI: 10.15888/j.cnki.csa.008474

      Abstract (383) HTML (740) PDF 1.73 M (840) Comment (0) Favorites

      Abstract:Machine-made sand is the fine aggregate for machine-made sand concrete. The quality of machine-made sand, determined by the stone powder content, has a significant impact on the strength, workability, durability, and other performance of machine-made sand concrete. Considering that with low accuracy and long duration, the traditional stone powder detection methods are cumbersome and difficult to quantify, this study proposes an improved UNet model based on the characteristics of machine-made sand. First, optical microscope equipment is used to collect images of machine-made sand particles, and these images are preprocessed by means of contrast enhancement, the look-up table algorithm, low-pass filtering, etc. Then, the deep residual and attention mechanism module is introduced to build an improved UNet model. Finally, segmentation and quantitative calculation are conducted on the stone powder in machine-made sand. The results show that the segmentation accuracy of the deep neural network constructed in this paper on the machine-made sand training dataset and the verification dataset is as high as 95.2% and 95.94%, respectively, and compared to the UNet, FCN, and Res-UNet methods, this method has significantly improved the segmentation effect on the same dataset.

    • Extraction of Pavement Potholes Based on Normal Vector Distance

      2022, 31(5):222-229. DOI: 10.15888/j.cnki.csa.008483

      Abstract (369) HTML (691) PDF 1.63 M (708) Comment (0) Favorites

      Abstract:Taking the laser point cloud of pavement elevation as the research object, this study proposes a method for extracting pavement potholes based on normal vector distance. Firstly, the cloud data of pavement elevation points are cleaned. Secondly, the PCA method in the adaptive optimal neighborhood is used to estimate the normal vector of the pavement point cloud data. The normal distance from the sampling point in the pavement point cloud to the tangent plane of its local quadric surface is calculated as the normal vector distance to describe the three-dimensional spatial features of the sampling points. Next, threshold segmentation is employed to automatically extract the pothole point cloud set, which is then segmented by the Mean-Shift clustering algorithm to obtain multiple pothole point sets. Finally, for each pothole point set, the Alpha Shape algorithm is used to extract pothole boundary points that are fitted by cubic spline interpolation three times to obtain the pothole contour. On this basis, the pothole size (length, width, and depth) and area are calculated. Experiments are carried out on regular pothole model point cloud data and real pavement point cloud data. The calculation shows that the average relative errors of pothole depth extracted by this method are 2.7% and 4.7%, respectively, and the average relative errors of pothole area extracted by this method are 6.8% and 4.3%, respectively. The experimental results show that the proposed method can accurately extract the boundary points and size information of pavement potholes and has a strong anti-interference ability for the recognition and extraction of irregular-shaped potholes.

    • Highway Road Pothole Extraction Method Based on RANSAC

      2022, 31(5):230-237. DOI: 10.15888/j.cnki.csa.008494

      Abstract (368) HTML (681) PDF 2.24 M (697) Comment (0) Favorites

      Abstract:To tackle the problems of heavy calculation burden and low efficiency of the pothole extraction algorithm based on the scanning of three-dimensional (3D) laser point clouds, this study proposes a pothole extraction method based on RANSAC. Firstly, RANSAC is employed to calculate the cross-sectional baseline for the correction of cross-sectional data and preliminary identification of pothole points and their locations. Secondly, the local reference road surface near the pothole is calculated by RANSAC so that the pothole points and road surface points can be marked. Thirdly, the seed filling algorithm is used to solve the connected domain and calculate the set of pothole points. Finally, the edge of the pothole is extracted with the set of pothole points and an exhaustive analysis of the pothole data is made. The experimental results show that RANSAC can quickly scan cross-sectional point cloud data, with the processing time increased by 56.46% on average compared with that of the curvature feature point detection algorithm. It has a good effect on extracting the depth and area of potholes with high accuracy. The average error of depth and area is 4.73% and 4.50%, respectively.

    • Short-term Traffic Flow Prediction Based on VMD-BiLSTM-BLS Model

      2022, 31(5):238-245. DOI: 10.15888/j.cnki.csa.008469

      Abstract (367) HTML (1414) PDF 1.68 M (900) Comment (0) Favorites

      Abstract:Accurate short-term traffic flow forecasting is very important in smart transportation systems. In recent years, bi-directional long-short term memory (BiLSTM) has been widely used in short-term traffic flow prediction, but due to its structural characteristics, it is prone to overfitting, affecting the prediction accuracy. Given that the broad learning system (BLS) can solve the problem of overfitting, this study combines deep learning with broad learning. Furthermore, the variational mode decomposition (VMD) is introduced for noise reduction so as to minimize the interference of noise on the traffic data. By doing this, the VMD-BiLSTM-BLS short-term traffic flow prediction model is proposed in this paper. The PeMS traffic flow data is used as an example for predictive analysis, and the results show that compared with the baseline model, the ablation model, and the existing model, the proposed model has the best prediction accuracy and can better reflect the short-term traffic flow at the intersection.

    • Generation Model from Structured Data to Numerical Analysis Text

      2022, 31(5):246-253. DOI: 10.15888/j.cnki.csa.008480

      Abstract (380) HTML (786) PDF 1.26 M (854) Comment (0) Favorites

      Abstract:Text generation based on structured data is an important research direction in the field of natural language generation. It can transform structured data collected by sensors or statistically analyzed by computers into natural language texts suitable for human reading and understanding, thus becoming an important technology for automatic report generation. It is of great application value to study models of generating texts from structured data for the generation of analytical texts from various types of numerical data in reports. In this paper, we propose an encoder-decoder text generation model incorporating the coarse-to-fine aligner selection mechanism and the linked-based attention mechanism, which matches the characteristics of numerical data, and consider the problems of excessive content dispersion and failure to highlight descriptions in the process of generating analytical texts from numerical data. In addition, we also model the relationship between the domains to which the numerical data specifically belong in order to improve the correctness of the discourse order in generated texts. Experimental results show that the model proposed in this paper, which incorporates both mechanisms, has better performance in terms of metrics than the traditional model based on the content-based attention mechanism only, the model based on both the content-based attention mechanism and the linked-based attention mechanism, and the GPT2-based model. This proves the effectiveness of the proposed model in the task of generating analytical texts with numerical data.

    • Prevention of Man-in-the-middle Attacks on BGP Using Certificateless Signatures

      2022, 31(5):254-261. DOI: 10.15888/j.cnki.csa.008531

      Abstract (436) HTML (436) PDF 1.05 M (642) Comment (0) Favorites

      Abstract:The border gateway protocol (BGP) is used to exchange network reachability information between autonomous systems, but it is threatened by man-in-the-middle attacks. Therefore, an improved certificateless multi-signature scheme is proposed and applied to BGP. The inter-domain routing must be signed according to the route delivery order, and the autonomous systems can receive the route only after the multi-signatures are verified successfully. The public and private keys to the autonomous systems are generated interactively with the trusted center with a fixed length of the signature message and efficient calculations. The security analysis proves that the proposed scheme cannot be fabricated under the random oracle model and is valid for resisting the man-in-the-middle attacks on BGP.

    • Solution to Backpressure Problem Based on Data Migration Strategy

      2022, 31(5):262-268. DOI: 10.15888/j.cnki.csa.008482

      Abstract (451) HTML (495) PDF 1.28 M (709) Comment (0) Favorites

      Abstract:In the application of stream computing, the mismatch of upstream and downstream data inflow and outflow speed often leads to the problem of insufficient data buffer capacity or overflow backpressure, and data loss and system crash are the possible consequences. A good solution is in urgent need. The existing methods address the downstream backpressure problem by transferring pressure upstream, but in this paper, a backpressure solution based on data migration strategy is proposed to solve the backpressure problem by dispersing the pressure to light-loaded nodes of other branches. The experiments on the NS-3 network simulation platform show that the proposed method has significantly improved the throughput proportion and latency in contrast to the Credit backpressure mechanism of the Flink framework.

    • Lossless and Efficient Threshold Color Image Secret Sharing and Information Hiding Algorithm Based on CRT

      2022, 31(5):269-276. DOI: 10.15888/j.cnki.csa.008429

      Abstract (635) HTML (861) PDF 1.47 M (812) Comment (0) Favorites

      Abstract:At present, the research on secret image sharing mainly focuses on gray-scale images. However, most used in daily life are color images. Therefore, it is of great significance and application value to study the secret sharing of color images. The scheme combines the secret sharing based on the Chinese remainder theorem (CRT) with DCT information hiding technology to ensure the security of color secret image transmission. At the generation end, the color secret shadow image generated by CRT is embedded into the color carrier image of users and distributed by DCT information hiding algorithm. The shadow image is extracted at the recovery end, and the color secret image is recovered by CRT. The process satisfies the (t, n) threshold. The experimental results show that the algorithm can achieve lossless recovery, and the evaluation with relevant parameters reveals that the scheme outperforms other ones.

    • Video Object Tracking Algorithm Based on Improved MDNet

      2022, 31(5):277-284. DOI: 10.15888/j.cnki.csa.008523

      Abstract (395) HTML (557) PDF 1.36 M (612) Comment (0) Favorites

      Abstract:One of the major problems of the object tracking algorithm is the imbalance of positive and negative samples, and the positive samples are of high similarity. Aiming at the problem of insufficient positive samples in the tracking update process, this study proposes an improved MDNet-based video object tracking algorithm based on the multi-domain convolutional neural network (MDNet) algorithm. First, the strategy of candidate selection is improved in the original algorithm, and a model update method is presented on the basis of the combination of the candidate confidence and the threshold judgment of coordinate variance. Second, the cross-entropy loss function of the original algorithm is altered to a focal loss function with better performance. The experimental results show that the algorithm has a significant improvement in tracking precision and success rate compared with the MDNet algorithm under the same experimental environment.

    • Recommendation Algorithm Combining Latent Factor Model and Gated Recurrent Unit

      2022, 31(5):285-290. DOI: 10.15888/j.cnki.csa.008455

      Abstract (393) HTML (477) PDF 1.26 M (647) Comment (0) Favorites

      Abstract:In traditional recommendation algorithms, there is often a lack of consideration of users’ long short-term interest preferences. However, with the deepening of the application of deep learning in recommendation algorithms, this problem can be solved well. In response to the problem, this study proposes a recommendation algorithm based on long short-term interest preferences (RA_LST), which integrates a latent factor model and a gated recurrent unit. It can capture users’ long short-term preferences respectively and thus effectively solves the problem that the recommendation effect decreases due to users’ interest changing with time. The final experimental results show that the proposed algorithm improves the recommendation accuracy on different data sets.

    • Criminal Investigation Image Super-resolution Based on DCT and Zero-shot Learning

      2022, 31(5):291-297. DOI: 10.15888/j.cnki.csa.008472

      Abstract (400) HTML (668) PDF 2.76 M (788) Comment (0) Favorites

      Abstract:Image super-resolution (SR) plays an important role in video-based criminal investigation. SR algorithms based on convolutional neural networks are usually trained with the input of artificially synthesized low-resolution images to learn the mapping between high-resolution and low-resolution images, and thus they are difficult to be applied in the field of video-based criminal investigation. The degradation process of real low-resolution images is complex and unknown, and most of them are processed by compression algorithms, leading to false textures in SR images. Therefore, a new SR algorithm based on discrete cosine transform (DCT) and zero-shot learning is proposed for the video-based criminal investigation images in real scenarios. The algorithm takes advantage of the repetitive similarity within the images and utilizes sub-images from the input image for training. Different from the input of the previous SR network, the proposed algorithm takes the DCT coefficients of the sub-images as the input of the SR network to avoid magnifying compression artifacts of the input image and reduce false textures. The experimental results on the standard datasets and real criminal investigation images show that the proposed algorithm can reduce the false texture caused by compression artifacts.

    • Improved Person Re-identification Based on Global Feature

      2022, 31(5):298-303. DOI: 10.15888/j.cnki.csa.008477

      Abstract (445) HTML (1394) PDF 1.17 M (898) Comment (0) Favorites

      Abstract:Person re-identification faces challenges such as posture change, occlusion interference, and illumination difference, and thus it is very important to extract pedestrian features with strong discriminability. In this paper, an improved person re-identification method based on global features is proposed. Firstly, a multi-receptive field fusion module is designed to fully obtain pedestrian context information and improve the global feature discriminability. Secondly, generalized mean (GeM) pooling is used to obtain fine-grained features. Finally, a multi-branch network is constructed, and the features of different depths of the network are fused to predict the identity of pedestrians. The mAP indexes of this method on Market1501 and DukeMTMC-ReID are 83.8% and 74.9%, respectively. The experimental results show that the proposed method can effectively improve the model based on global features and raise the recognition accuracy of person re-identification.

    • Multi-round Verification Scheme Using State Reduction to Process Cross-shard Transactions

      2022, 31(5):304-315. DOI: 10.15888/j.cnki.csa.008485

      Abstract (315) HTML (435) PDF 1.89 M (694) Comment (0) Favorites

      Abstract:In the Blockchain system, sharding is the main on-chain expansion solution, and state sharding can solve the scalability problem of the public chain without reducing security. However, the introduction of sharding technology has also brought in the processing problem of cross-shard transaction verification. When most transactions in the system are cross-shard transactions, the ability of processing cross-shard transactions determines the performance of the entire system. Therefore, cross-sharding transaction verification and processing strategies are very important in the process of designing the sharding system. In response to the above problems, this study proposes a state reduction and multi-round (SRMR) scheme that uses state reduction combined with multiple rounds of verification within shards to process cross-shard transactions. First, the probability of cross-shard transactions is analyzed, and then the probability of cross-shard transaction processing in each layer is evaluated under the proposed model of using state reduction to process cross-shard transactions. It is found that the state reduction model alone will make the upper-layer shard transaction load unduly large. Thus, the incentive mechanism and the state reduction combined with multi-round verification are put forward to balance the upper-layer transaction load. Finally, the value of reasonable rounds is obtained, and a strategy of reasonably balancing reduction and multi-round verification is presented. This scheme comprehensively utilizes the capabilities of nodes to ensure the smooth completion of cross-shard transactions and reduce the rollback per cross-shard transaction.

    • Target Recognition in Urban Remote Sensing Images Based on Multi-feature Space and Its Optimization

      2022, 31(5):316-323. DOI: 10.15888/j.cnki.csa.008458

      Abstract (529) HTML (553) PDF 1.87 M (904) Comment (0) Favorites

      Abstract:Target recognition in urban remote sensing images can help monitor the types of urban features and is a hot research topic in recent years. However, the traditional pixel-based method cannot make full use of the features of high-resolution remote sensing images, whereas the traditional object-based method cannot accurately extract the objects. To address the shortcomings of the traditional methods, this study proposes a method of target recognition in urban remote sensing images based on the multi-feature space and its optimization. This method takes the two traditional methods as the premise, combines pixel features with object features, and constructs the multi-feature space by supplementing depth features provided by the VGG19 network. The XGBoost algorithm is used to select features in the multi-feature space. An optimal feature space is established and sent to the random forest recognizer to achieve the target recognition in urban remote sensing images. The experimental results show that the recognition accuracy of the proposed method is 87.89%, and the Kappa coefficient is 0.83, which means this method displays a high recognition capability in the study area and is an effective method for target recognition in urban remote sensing images.

    • Image Super-resolution Reconstruction Based on Multi-scale Attention Residual Network

      2022, 31(5):324-330. DOI: 10.15888/j.cnki.csa.008465

      Abstract (570) HTML (536) PDF 1.31 M (943) Comment (0) Favorites

      Abstract:Digital images play an important role in information transmission, and image super-resolution technology can enrich image details. To address the problems of insufficient effective feature reuse of low-resolution images and excessive parameters in many networks, this study combines convolution kernels of different sizes and attention residual mechanism to construct the image super-resolution network. Three convolution layers of different scales are used to extract the image features, of which the second and third layers replace the large convolution kernels with small ones, and after the three-layer convolution fusion, the attention mechanism is introduced. Finally, the traditional Bicubic interpolation is used to directly provide low-frequency information for the network. By doing this, while reducing the number of parameters and mitigating the disappearance of gradients, the proposed network can make the effective high-frequency information gain greater weights and can enhance the nonlinear expression ability between the networks, which is conducive to the iterative convergence of network training. Experimental results show that the proposed network can enhance the image reconstruction ability to a certain extent.

    • Detection of Pin Missing from Nuts of Transmission Tower Based on Federated Learning

      2022, 31(5):331-337. DOI: 10.15888/j.cnki.csa.008479

      Abstract (526) HTML (681) PDF 1.42 M (808) Comment (0) Favorites

      Abstract:The nut on the transmission tower is the medium connecting two or more transmission tower components, and the pin is an important guarantee to ensure that the nut does not fall off. The lack of pins will lead to potential safety hazards at the joints between various components. This study combines the federated learning and target detection algorithm to upload the local model and generate the fusion model through the central node without any data exchange among regions. The detection algorithm Faster RCNN and the classification network are used to detect and classify nuts, respectively. The experimental results show that compared with local models, the fusion model based on federated learning improves the mAP of detection tasks by 3%–6% and the accuracy of classification tasks by 2%–3%.

    • Anomaly Identification Method for Pavement Sensor Monitoring Data

      2022, 31(5):338-344. DOI: 10.15888/j.cnki.csa.008484

      Abstract (318) HTML (576) PDF 1.56 M (682) Comment (0) Favorites

      Abstract:The full-scale accelerated loading test field has a complex pavement structure, in which a variety of sensors are embedded to monitor indicators of pavement performance. For the high-frequency and massive data collected by the sensors, the identification of abnormal data using traditional methods has low efficiency and poor accuracy. Considering this, this study visualizes the originally collected high-frequency data through specific software and then labels the visualized data as the original dataset. Next, according to the characteristics of obvious shape features of the data after visualization, the lightweight convolutional neural network model GhostNet is selected to automatically identify the abnormal data from the monitored dataset by sensors. Through the parameter design and the network model training, the test results on the verification set show that the identification rate of abnormal data is as high as 99%. Compared with the conventional classification model residual neural network (Resnet50), the GhostNet model has improved the anomaly identification accuracy by 11%. It can quickly identify abnormal data in massive monitored data by pavement sensors, which can provide strong data support for pavement sensor fault monitoring.

    • Association Analysis of College Physical Fitness Test Data Based on Improved Apriori Algorithm

      2022, 31(5):345-350. DOI: 10.15888/j.cnki.csa.008447

      Abstract (346) HTML (658) PDF 1.06 M (802) Comment (0) Favorites

      Abstract:To effectively analyze college physical fitness test data and quickly feed back the factors that affect students’ test results, this study takes the physical fitness test data of the North University of China as the sample and transforms preprocessed data into datasets suitable for data mining. Considering the limited features and consistent length of physical fitness test data, an Apriori algorithm that combines the transaction reduction technique with the hash technique is used to analyze data, which reduces the number of database traversal and the scale of candidate sets generated. It also improves the efficiency of the algorithm and ensures mining accuracy at the same time. Finally, comparison and analysis are made with the Apriori algorithm, the Apriori algorithm based on transaction reduction, and the Apriori algorithm based on the hash technique. The experimental results show that the proposed improved Apriori algorithm that combines transaction reduction and the hash technique can effectively analyze the association rules among students’ physical fitness test results and therefore has a stronger guiding significance for students’ physical fitness training. Compared with the Apriori algorithm, the proposed algorithm improves the operation efficiency by more than 85%.

    • Conversion Method from Fortran to CUDA C

      2022, 31(5):351-357. DOI: 10.15888/j.cnki.csa.008449

      Abstract (346) HTML (920) PDF 1.09 M (742) Comment (0) Favorites

      Abstract:Graphic processing unit (GPU)-based heterogeneous computing has gradually become the mainstream computing method. Nevertheless, due to the limited historical development of scientific computing programming, a lot of numerical computing software is still implemented in Fortran. In terms of increasing the computing speed, a large amount of software needs to be transplanted onto compute unified device architecture (CUDA) C. However, it would be a complicated and massive project to manually implement the program transplant. If automatic conversion from Fortran to CUDA C can be achieved, the efficiency of program development would be greatly improved. This study designs an algorithm converting Fortran to CUDA C, implements the algorithm through regular expressions and shell scripts, and verifies it by programming test cases. Experimental results show that this tool is reliable, stable, and compatible. In the transplant process of large programs, it can automatically filter and establish variable information tables and generate CUDA-related operation functions. The resulting code possesses good readability, and the conversion accuracy is more than 80%. The workload of the transplant is effectively reduced.

    • Foreign Object Detection of Coal Mine Belt Conveyor Based on Deep Generative Model

      2022, 31(5):358-363. DOI: 10.15888/j.cnki.csa.008468

      Abstract (497) HTML (564) PDF 1.07 M (1424) Comment (0) Favorites

      Abstract:A foreign object detection method based on the deep generative model is proposed to accurately detect the foreign objects on the coal mine belt conveyor. First, a conventional variational auto-encoder (VAE) is used to reconstruct the image, and the presence of foreign objects in the image is detected according to the reconstruction error between the original image and the reconstructed image. Considering that the reconstructed image generated by the VAE is usually fuzzy, a generative adversarial network (GAN) is introduced to evaluate the original image and the reconstructed image for a clearer image and higher foreign object detection accuracy. Finally, the VAE is combined with the GAN to design a deep learning algorithm suitable for belt foreign object detection. The experimental results show that compared with the baseline method the proposed method has a better effect on every evaluation indexes.

    • Application of Improved DBSCAN Algorithm in Similarity of Campus Trajectory Data

      2022, 31(5):364-370. DOI: 10.15888/j.cnki.csa.008486

      Abstract (326) HTML (1011) PDF 1.27 M (881) Comment (0) Favorites

      Abstract:An improved DBSCAN spatiotemporal clustering algorithm is proposed to increase the intimacy analysis accuracy of social relationships hidden in campus wireless network data. First, spatiotemporal trajectories are formed according to the location and time of the WiFi connection by collecting campus wireless network data, and an improved algorithm is used to classify the spatiotemporal trajectories. Then, the characteristic trajectories of the clustering results are extracted, and the LCSS algorithm is employed to measure the similarity of spatiotemporal trajectories. The high similarity between the trajectories indicates the close relationships, and the low similarity reveals those isolated students that need to be further investigated and counseled by teachers. Finally, FinBI is used to visualize the trajectory clustering results. The experimental results show that the improved algorithm can increase the accuracy and effectiveness of the clustering results while providing a reference for solving other similarity problems.

    • Self-supervised Multi-view Subspace Clustering with Graph Information

      2022, 31(5):377-381. DOI: 10.15888/j.cnki.csa.008448

      Abstract (318) HTML (680) PDF 863.37 K (730) Comment (0) Favorites

      Abstract:Multi-view subspace clustering methods are usually used to process high-dimensional and complex data. Most of the existing multi-view subspace clustering methods analyze and process data by mining potential graph information, with no supervision process for the representation of the potential subspace. To solve this problem, this study proposes a new multi-view subspace clustering method, namely self-supervised multi-view subspace clustering (SMSC) based on graph information. It combines spectral clustering with subspace representation to formulate a unified deep learning framework. SMSC constructs potential graph information by mining the first-order and second-order graphs of multi-view data and then uses clustering results to supervise the learning process of the common potential subspace of multi-view data. Extensive experiments on four standard datasets show that the proposed method is more effective than traditional multi-view subspace clustering methods.

    • Digital Image Hiding Technology Combining 2D Logistic Mapping and 2D-DCT

      2022, 31(5):382-387. DOI: 10.15888/j.cnki.csa.008495

      Abstract (365) HTML (482) PDF 1.15 M (634) Comment (0) Favorites

      Abstract:To cope with increasingly serious problems related to image security, this study proposes a digital image hiding algorithm based on two-dimensional (2D) Logistic mapping and 2D discrete cosine transform (2D-DCT). Firstly, the chaotic sequence generated by the 2D Logistic mapping is used to diffuse and scramble pixels of the secret image so that the secret image can be encrypted. Then, 2D-DCT is performed on the host image in blocks, and the image information after diffusion and scrambling is stored in the lower right corner of each block after 2D-DCT. Finally, 2D inverse discrete cosine transform (2D-IDCT) is carried out to yield the stego image. Experimental results also show that the proposed algorithm is safe, feasible, and effective in image hiding.

Current Issue


Volume , No.

Table of Contents

Archive

Volume

Issue

联系方式
  • 《计算机系统应用》
  • 1992年创刊
  • 主办单位:中国科学院软件研究所
  • 邮编:100190
  • 电话:010-62661041
  • 电子邮箱:csa (a) iscas.ac.cn
  • 网址:http://www.c-s-a.org.cn
  • 刊号:ISSN 1003-3254
  • CN 11-2854/TP
  • 国内定价:50元
You are the firstVisitors
Copyright: Institute of Software, Chinese Academy of Sciences Beijing ICP No. 05046678-3
Address:4# South Fourth Street, Zhongguancun,Haidian, Beijing,Postal Code:100190
Phone:010-62661041 Fax: Email:csa (a) iscas.ac.cn
Technical Support:Beijing Qinyun Technology Development Co., Ltd.

Beijing Public Network Security No. 11040202500063