首页期刊导航|Journal of Electronic Imaging
期刊信息/Journal information
Journal of Electronic Imaging
International Society for Optical Engineering
Journal of Electronic Imaging

International Society for Optical Engineering

双月刊

1017-9909

Journal of Electronic Imaging/Journal Journal of Electronic ImagingSCI
正式出版
收录年代

    Systematic review and analysis on underwater image enhancement methods, datasets, and evaluation metrics

    Gunjan VermaManoj Kumar
    060901.1-060901-31页
    查看更多>>摘要：Underwater environments can be used to explore new resources that can beemployed in the fields of medical science and energy resources. Humans are dependent on thevaluable resources that exist beneath Earth’s surface. Underwater exploration requires enhancedimages that are obtained using enhancement methods. So, it is important that underwater imageenhancement (UIE) methods work well in terms of performance and accuracy. As a result,research in UIE has increased in the past few years. An extensive survey is conducted on existingUIE methods along with their broad classification, underwater datasets, and evaluation metrics,respectively. The experimental analysis is conducted to compare the existing UIE methods interms of qualitative and quantitative evaluation metrics. The real-world applications and futurescope of existing enhancement methods are highlighted and discussed.
      原文链接:
    • NETL
    • NSTL
    • Spie-Soc Photo-Optical Instrumentation Engineers

    Secure framework against cyber attacks on cyber-physical robotic systems

    Akashdeep BhardwajMohammad Dahman AlshehriKeshav KaushikHasan J. Alyamani...
    061802.1-061802.21页
    查看更多>>摘要：Robot-based platforms and processes have integrated the security and efficiency ofdata into a comprehensive range for domains, such as manufacturing, industrial, logistical, agricultural,healthcare, and internet services. Smart cyberattacks have been on the rise, specificallytargeting corporate, industrial robotic systems. These attacks execute once the internet of things,internet, and organization integration is implemented with the industrial units. We implementedsecurity criteria-based indices for cyber-physical systems (CPS) with industrial components andembedded sensors that process the information logs and processes. We proposed an attack treebasedsecure framework that does not include every CPS device but takes into consideration thecritical exploitable vulnerabilities to execute the attacks. We categorized each physical deviceand integrated sensors based on logs and information in a sensor indices device library. Thisresearch simulated real-time exploitation of vulnerabilities on CPS robotic systems using theproposed framework in form of a two-phased process. This validates the enhanced data securityoutput of the integrated sensor and physical nodes with the intelligent monitor and controllersystem health monitor during real-time cyberattacks This research simulated common cyberattackson cyber-physical controller servers based on cross-site scripting and telnet pivoting. Theauthors gathered known and unknown vulnerabilities and exploited them with a tree-based attackalgorithm. The authors calculated the average time for cyberattackers with different skills whentrying to compromise CPS devices and systems.
      原文链接:
    • NETL
    • NSTL
    • Spie-Soc Photo-Optical Instrumentation Engineers

    Target recognition and localization based on lightweight single-shot multibox detector network for robotics

    Hua ZhangMingjian NiuXiaochu ChenJie Wu...
    061803.1-061803.10页
    查看更多>>摘要：Target recognition and localization are essential in computer vision and pattern recognitionin robotics. The artificial extraction of features was omitted with the emergence of deepconvolutional neural networks, reducing the influence of human factors on the results. The single-shot multibox detector (SSD) network has achieved excellent recognition by high precisionand fast speed in target recognition and positioning. However, some small real-time systems aredifficult to implement because of their demanding hardware and extended training time. Basedon the digital normalization and residual network structure, the depth-wise separable convolutionis proposed to replace the traditional convolution. The improved SSD network structure was usedfor identification and positioning in our work. The speed of training and testing increased withouta decline in the accuracy, thus reducing the dependence on hardware. The method hasachieved good results on the PASCAL VOC dataset after testing. It can also be applied to thefield of intelligent inspection robots and intelligent security robots.
      原文链接:
    • NETL
    • NSTL
    • Spie-Soc Photo-Optical Instrumentation Engineers

    Improved face detection algorithm based on multitask convolutional neural network for unmanned aerial vehicles view

    Zhihe XunLiang WangYunqing Liu
    061804.1-061804.10页
    查看更多>>摘要：With the rapid development of automation technology, unmanned aerial vehicles(UAV) have been widely used to assist troops in long-distance combat missions. However, consideringthat the UAV needs to keep flying and maintain a certain distance from the target, facedetection using UAV usually achieves poor results, which hinders the application of the facerecognition technology in UAV systems.We proposed an improved Mish-L2-multitask convolutionalneural network (MTCNN) model based on MTCNN model to further improve the accuracyof small-size face detection. First, the maximum pooling layer of the P-Net CNN wasremoved. Second, a regularization term was added to the crossentropy loss function. Finally, theactivation functions exerted in three subneural networks of MTCNN model were replaced withMish activation functions. The result shows that the proposed Mish-L2-MTCNN model couldimprove the accuracy of small-size face detection efficiently under the UAV view. Comparingwith the result obtained from the original MTCNN model, the accuracy was improved by 3.62%,and the missing rate was evidently reduced. This work can provide methodological guidance forthe development of a UAV-based face recognition system in the real airborne scenario and guaranteethe validity and efficiency of the system. The study can also guide further research concerningthe detection effect in the cases of side face and half mask.
      原文链接:
    • NETL
    • NSTL
    • Spie-Soc Photo-Optical Instrumentation Engineers

    Temporary Petri-nets-based method for synthesizing network models

    Amin Salih MohammedAndriy KovalenkoNina Kuchuk
    061805.1-061805.15页
    查看更多>>摘要：Many legacy systems are still essential. They run efficiently and accurately in legacyhardware and software environments. Therefore, it is necessary to migrate legacy hardware andsoftware to modern platforms. One of the directions of migration is to service-oriented architectures.In this case, side effects are possible. Therefore, it is necessary to use the softwaresystem model. This model is used to assess the migration process of legacy software systems.An approach for building network models from the ground up is proposed in this research. TimedependentPetri nets are used to explain the traces of events in the simulated network. A user maycreate a network workload model that can be utilized for network planning purposes using thistechnique. A method must be established to determine whether or not the created model is suitable.An example is used to determine the suitability of the Markov model for the behavior of asystem with discrete states in terms of precision. An example is used to build a model for theoperation of a basic software system. Providing the executable network model of the migratingsystem to the service identification approach input is preferred.
      原文链接:
    • NETL
    • NSTL
    • Spie-Soc Photo-Optical Instrumentation Engineers

    Small obstacles detection on roads scenes using semantic segmentation for the safe navigation of autonomous vehicles

    Sadaf YasminMehr Yahya DurraniSaira GillaniMaryam Bukhari...
    061806.1-061806-23页
    查看更多>>摘要：Currently, autonomous vehicles and intelligent robots are used in an extensive set ofindustrial applications. However, in the recent past, the frequency with which small obstaclesarise on highways has risen dramatically, which will lead to severe incidents on highways.Hence, small obstacle detection is critical for increasing the efficiency of autonomous vehiclesto enable their safe navigation to avoid road accidents. On the other hand, these small-sizeobstacles are of varying sizes, shapes, and colors and are difficult to be detected under low lightingand illumination conditions. Existing research studies suggest deep-learning-assisted semanticsegmentation models; however, an optimal model along with improved performance is ofgreater necessity. From this perspective, we have suggested transfer learning-based approachesusing state-of-the-art semantic segmentation models namely UNet++, PSPNet, PANNet,LinkNet, and DeepLabV3+ for the detection of small-size obstacles under strict lighting andillumination conditions. Furthermore, we used images of road scenes with extremely small-sizeobstacles, which were neglected by past research studies, since these obstacles were not includedin the databases that they were using in their findings and may have failed to effectively addressthe problem of obstacle detection. Moreover, for faster and better convergence, we have modifiedthe backbone architectures of these models with Residual-Network (ResNet)-18 andResNet-34 trained on ImageNet weights. It is observed that DeepLabV3 + ResNet-18 as backbonearchitecture shows the highest results with a mean intersection-over-union score of 64%along with a 95% value of accuracy.
      原文链接:
    • NETL
    • NSTL
    • Spie-Soc Photo-Optical Instrumentation Engineers

    Consumer shopping emotion and interest database: a unique database with a multimodal emotion recognition method for retail service robots to infer consumer shopping intentions better than humans

    Xiaoyu ChenShouqian SunZiqi ZhangZirui Ma...
    061807.1-061807.19页
    查看更多>>摘要：Empowering retail service robots with empathy is one of the current research hotspotsin the field of artificial intelligence. Identifying consumer emotions, understanding thechanges in shopping interests, and developing appropriate sales strategies is a challenging taskfor retail service robots. We investigate the feasibility of using computer vision methods forempowering robots with empathy by examining the correlation between consumer emotion andlevels of shopping interest. To this end, we construct the first video database of consumer sentimentchanges in a business context and propose a deep learning method that uses multimodalinformation to infer consumers’ shopping intentions, and conduct preliminary experimental validationon this database. The experimental results show that the proposed method is 7% and 10%more accurate than manual assessment (n ¼ 40) in identifying consumer emotions and predictingconsumer shopping interest levels, respectively. Thus, the proposed method is valid andeffective. We anticipate that the results of this study will have considerable implications forhuman–computer interaction research in service robots.
      原文链接:
    • NETL
    • NSTL
    • Spie-Soc Photo-Optical Instrumentation Engineers

    Robust encrypted face recognition robot based on bit slicing and Fourier transform for cloud environments

    Milos AntonijevicIvana StrumbergerSasa LazarevicNebojsa Bacanin...
    061808.1-061808.13页
    查看更多>>摘要：In the field of computer vision, face recognition has become a trending researchtopic, and is widely used in the area of network security. The transmission of data over a networkvia a cloud server exposes the information to security risks and privacy attacks, meaning that theuse of an encryption algorithm is essential. Face recognition algorithms in robotics applicationshave become cumbersome in terms of the computation speed needed to recognize the image.Since this is a built-in programming function in the robotics board, it will limit the speed andsecurity of data storage. To overcome this issue, a cloud server is utilized as this can improve theprocessing speed, throughput, efficiency, and robustness of face recognition. Storing images inthe cloud server in a secure way requires that the image be encrypted. To achieve this, we proposea hybrid encryption technique based on bit slicing and a discrete Fourier transform (DFT),and develop a secure robotic face image recognition scheme using a MobileFaceNet-CNN model(BS-DFT-MobileFaceNet). The percentages of improvement in terms of accuracy over LeNet,VGG16Net, Alexnet, GoogLeNet, ResNet, MobileFaceNet including raw input image were17.5%, 7.36%, 22.41%, 22.62%, 8.65%, and 288.89% for raw input images and encryptedimages using Genetic Algorithm (GA), DNA Algorithm, Bit slicing, AES, DFT, and BS-DFTencryption algorithms respectively.
      原文链接:
    • NETL
    • NSTL
    • Spie-Soc Photo-Optical Instrumentation Engineers

    Markov decision process with deep reinforcement learning for robotics data offloading in cloud network

    Roobaea AlroobaeaAhmed BinmahfoudhSabah M. AlzahraniAnas Althobaitic...
    061809.1-061809.12页
    查看更多>>摘要：Robots have a wide range of computer capacities, and executing powerful computationalprograms on them might be difficult due to restricted internal processing, memory, andenergy. Similarly, cloud computing enables on-demand computation, so integrating robotics andcloud computing can help robots solve limitations. The key to successfully offloading jobs is anoperational solution that would not underutilize the robot’s natural processing capacity and actauthorized based on important costing criteria like delay and CPU resources. Applications areoffloaded from robots based on the Markovian decision process. The Markovian decision helpsto identify the resources in the cloud network based on probability. A deep reinforcement learning-based deep Q-network (DQN) technique selects resources in the cloud network. Further dataare offloaded in cloud storage. The state-space is built on the notion that the size of the inputinformation has a serious influence on the software’s processing time. The suggested techniqueis constructed as a repetitive work issue with a distinct space domain, in which we take a differentactivity at every successive stage using the resulting result to train the DQN to get the mostprizes. A navigation testbed was created and implemented to validate the suggested method.The proposed method minimizes the cost of communications between clouds and also itminimizes the latency of the application. It increases the accuracy level by 85%, which ishigher when compared with existing methods.
      原文链接:
    • NETL
    • NSTL
    • Spie-Soc Photo-Optical Instrumentation Engineers

    Target recognition approach using image local features in rehabilitation robots

    Milos AntonijevicDijana JovanovicSasa LazarevicDjordje Mladenovic...
    061810.1-061810.13页
    查看更多>>摘要：From the computer science literature, it can be seen that many different technologiesare used in target recognition, which is one of the most significant areas in the artificial intelligencefield. Target recognition is applied in a variety of disciplines, including healthcare, robotvision, vehicular traffic, and virtual reality. Target recognition techniques involve a robotic visionsystem that must perform with high accuracy and efficiency in real time; additionally, it musthave the capacity to handle difficult identification contexts. In one existing target recognitionsystem, the Harris algorithm is used; it provides a higher accuracy compared to more traditionalalgorithms. In order to improve its achieved accuracy, we focus on the target detection algorithmof a rehabilitation robot that is based on the local features of images. Considering the featurepoints of the images and target identification technology, a rehabilitation robotic recognitionmethod is developed in this work. Initially, it collects the images, and then, adaptive weightedsymplectic geometry decomposition is used for pre-processing. This method helps to reduce thenoise in the images. Next, the features are extracted, and the vectors of the features are separatedand identified. Afterward, one-to-many rehabilitation modes and actual system monitors areimplemented to precisely select the target condition based on the functional criteria of the rehabilitationrobot recognition method. Finally, an invertible color-to-grayscale conversion methodusing clustering and reversible watermarking is applied. It converts images into grayscale. TheGaussian distribution is consistently utilized to define the position and the quantity of theextracted feature points. Related images are retrieved as well. According to experimental findings,the proposed method improves the accuracy and the recall rate compared with the Harrisalgorithm.
      原文链接:
    • NETL
    • NSTL
    • Spie-Soc Photo-Optical Instrumentation Engineers