Pub Date : 2021-12-10DOI: 10.1109/ICCSS53909.2021.9721951
S. Li, Xinyu Li, Y. Zuo, Tie-shan Li
Predicting the fuel consumption of ships sailing under different navigation conditions and improving the operation efficiency of shipping industry has become an important topic. There are many characteristic variables affecting ship fuel consumption during navigation, such as trim, draft, wind speed, wind direction and so on. And some variables are highly correlated, which is easy to produce multicollinearity problems. It makes the fuel consumption prediction complex. The study established an Elastic network regression model by combining the least absolute contraction and selection operator (LASSO) and Ridge regression algorithm. The model reduces the complexity and improves the interpretability and accuracy by selecting the characteristic variables affecting ship fuel consumption. The study is verified by the navigation data of a ferry within two months. The results show that compared with long short term memory (LSTM) and back-propagation neural network (BPNN), the Elastic network regression model can not only explain the relationship between fuel consumption and variables, but also predict fuel consumption more accurately and effectively.
{"title":"Prediction of ship fuel consumption based on Elastic network regression model","authors":"S. Li, Xinyu Li, Y. Zuo, Tie-shan Li","doi":"10.1109/ICCSS53909.2021.9721951","DOIUrl":"https://doi.org/10.1109/ICCSS53909.2021.9721951","url":null,"abstract":"Predicting the fuel consumption of ships sailing under different navigation conditions and improving the operation efficiency of shipping industry has become an important topic. There are many characteristic variables affecting ship fuel consumption during navigation, such as trim, draft, wind speed, wind direction and so on. And some variables are highly correlated, which is easy to produce multicollinearity problems. It makes the fuel consumption prediction complex. The study established an Elastic network regression model by combining the least absolute contraction and selection operator (LASSO) and Ridge regression algorithm. The model reduces the complexity and improves the interpretability and accuracy by selecting the characteristic variables affecting ship fuel consumption. The study is verified by the navigation data of a ferry within two months. The results show that compared with long short term memory (LSTM) and back-propagation neural network (BPNN), the Elastic network regression model can not only explain the relationship between fuel consumption and variables, but also predict fuel consumption more accurately and effectively.","PeriodicalId":435816,"journal":{"name":"2021 8th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS)","volume":"51 8","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114105535","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-10DOI: 10.1109/ICCSS53909.2021.9721984
Jing Han, Guici Chen, Guodong Zhang, Junhao Hu
In this paper, a class of delayed inertial neural networks(INNs) with fuzzy templates is considered. By using a novel Lyapunov functional and an effective control law, several synchronization results of the investigated delayed fuzzy inertial neural networks(FINNs) are given. We construct the exponential synchronization results of the delayed FINNs via the nonreduced-order method for the first time. At last, numerical simulations show the correctness of the obtained results.
{"title":"Synchronization control for a class of delayed fuzzy inertial neural networks","authors":"Jing Han, Guici Chen, Guodong Zhang, Junhao Hu","doi":"10.1109/ICCSS53909.2021.9721984","DOIUrl":"https://doi.org/10.1109/ICCSS53909.2021.9721984","url":null,"abstract":"In this paper, a class of delayed inertial neural networks(INNs) with fuzzy templates is considered. By using a novel Lyapunov functional and an effective control law, several synchronization results of the investigated delayed fuzzy inertial neural networks(FINNs) are given. We construct the exponential synchronization results of the delayed FINNs via the nonreduced-order method for the first time. At last, numerical simulations show the correctness of the obtained results.","PeriodicalId":435816,"journal":{"name":"2021 8th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125137746","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-10DOI: 10.1109/ICCSS53909.2021.9722034
Zhuangbi Lin, Zhi Liu
The adaptive consensus control for multi-agent systems (MASs) with actuator failures is considered in this article. By combining the neural networks (NNs) technique to develop the control scheme, the unknown nonlinear function are allowed to exist in the system dynamics. Moreover, the disturbance is also compensated by adaptive estimated parameter. The controller is totally distributed and only two unknown parameters needed to be updated. The presented control method not only ensures that every agent of MAS can track the leader with a predetermined error, but improves the transient performance. At last, a physical example is provided to demonstrate the effectiveness of the proposed method.
{"title":"Adaptive Neural Consensus Control of Nonlinear Multi-agent Systems with Actuator Failures","authors":"Zhuangbi Lin, Zhi Liu","doi":"10.1109/ICCSS53909.2021.9722034","DOIUrl":"https://doi.org/10.1109/ICCSS53909.2021.9722034","url":null,"abstract":"The adaptive consensus control for multi-agent systems (MASs) with actuator failures is considered in this article. By combining the neural networks (NNs) technique to develop the control scheme, the unknown nonlinear function are allowed to exist in the system dynamics. Moreover, the disturbance is also compensated by adaptive estimated parameter. The controller is totally distributed and only two unknown parameters needed to be updated. The presented control method not only ensures that every agent of MAS can track the leader with a predetermined error, but improves the transient performance. At last, a physical example is provided to demonstrate the effectiveness of the proposed method.","PeriodicalId":435816,"journal":{"name":"2021 8th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125164728","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-10DOI: 10.1109/ICCSS53909.2021.9722007
Duxin Chen, Wenwu Yu, Qi Shao, Xiaolu Liu
Various real world data contains complex coupling spatio-temporal information, which brings a huge challenge for prediction, especially long-term prediction. Therefore, in this study, we propose a causality induced spatiotemporal feature extraction method and a novel deep learning framework for long-term strongly coupling data prediction tasks, which can effectively extract long-term spatio-temporal dependence of the time series data through causal network, geographic network and multiple time extraction mechanism. The proposed algorithm has achieved outstanding prediction performance in the widely- used test data set of traffic flow, where the long-term prediction accuracy of is nearly 30% better than other state-of-the-art currently-used spatio-temporal prediction models.
{"title":"Causality Induced Distributed Spatio-temporal Feature Extraction","authors":"Duxin Chen, Wenwu Yu, Qi Shao, Xiaolu Liu","doi":"10.1109/ICCSS53909.2021.9722007","DOIUrl":"https://doi.org/10.1109/ICCSS53909.2021.9722007","url":null,"abstract":"Various real world data contains complex coupling spatio-temporal information, which brings a huge challenge for prediction, especially long-term prediction. Therefore, in this study, we propose a causality induced spatiotemporal feature extraction method and a novel deep learning framework for long-term strongly coupling data prediction tasks, which can effectively extract long-term spatio-temporal dependence of the time series data through causal network, geographic network and multiple time extraction mechanism. The proposed algorithm has achieved outstanding prediction performance in the widely- used test data set of traffic flow, where the long-term prediction accuracy of is nearly 30% better than other state-of-the-art currently-used spatio-temporal prediction models.","PeriodicalId":435816,"journal":{"name":"2021 8th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126652028","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-10DOI: 10.1109/ICCSS53909.2021.9722008
Lequan Wang, Jin Duali, Ziqiang Chen, Guangqiu Chen, Gaotian Liu
Over-fitting is a severe problem when we adopt deep neural networks with a large number parameters in fine-grained visual classification. Many data augmentation methods are proposed through weakly supervised learning to alleviate over-fitting issue. Different from those methods, we propose a weakly supervised attention-guided regularization by object parts’ attention maps to fine-tune the Fully Connected (FC) layer and relieve over-fitting issue during training in this paper. On the other hand, the neural units in the last convolutional layer contain the same receptive fields that limit recognition performance due to involving lots of background noises. To alleviate this issue, we devise a spatial information mining module with an auxiliary penalty loss to aggregate multi-scale receptive fields feature maps with the selected precedent layer. Comprehensive experiments are conducted to show our method achieves or surpasses state-of-the-art results on common fine-grained classification datasets.
{"title":"Weakly Supervised Fine-Grained Visual Classification Through Spatial Information Mining and Attention-guided Regularization","authors":"Lequan Wang, Jin Duali, Ziqiang Chen, Guangqiu Chen, Gaotian Liu","doi":"10.1109/ICCSS53909.2021.9722008","DOIUrl":"https://doi.org/10.1109/ICCSS53909.2021.9722008","url":null,"abstract":"Over-fitting is a severe problem when we adopt deep neural networks with a large number parameters in fine-grained visual classification. Many data augmentation methods are proposed through weakly supervised learning to alleviate over-fitting issue. Different from those methods, we propose a weakly supervised attention-guided regularization by object parts’ attention maps to fine-tune the Fully Connected (FC) layer and relieve over-fitting issue during training in this paper. On the other hand, the neural units in the last convolutional layer contain the same receptive fields that limit recognition performance due to involving lots of background noises. To alleviate this issue, we devise a spatial information mining module with an auxiliary penalty loss to aggregate multi-scale receptive fields feature maps with the selected precedent layer. Comprehensive experiments are conducted to show our method achieves or surpasses state-of-the-art results on common fine-grained classification datasets.","PeriodicalId":435816,"journal":{"name":"2021 8th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114493022","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-10DOI: 10.1109/ICCSS53909.2021.9721950
Shushi Liu, Chuandong Li
This paper presents a new method to minimize the error function between the expected spike time and the actual spike time, which is a parallel combination of facilitating synapses consisting of an excitatory and an inhibitory synapse. The SpikeProp algorithm is designed to solve the error optimization problem between the expected spike time and the actual spike time of the current from the presynaptic neuron passing through the synapse to the postsynaptic neuron. The SpikeProp algorithm merges the Bienenstock–Cooper–Munro (BCM) rule with Spike Timing Dependent Plasticity (STDP) before calculating errors. The idea of filtration based on value in Synaptic Weight Association Training (SWAT) is utilized in the hidden layer. Thus, a time selector is used in the synapse between the input layer and the hidden layer, which is achieved through parallel combination of excitatory and inhibitory synapses. The neuron models used in these two processes are Leaky Integrate and Fired (LIF) and Spike Response Model (SRM), respectively. The algorithm is benchmarked against the nonlinear exclusive OR (XOR) problem. The simulation results has illustrated the diagram of the time selector in the hidden layer and the error measured in the output layer.
本文提出了一种最小化预期尖峰时间与实际尖峰时间之间误差函数的新方法，该方法是由兴奋性突触和抑制性突触组成的便利突触并行组合。SpikeProp算法旨在解决从突触前神经元到突触后神经元的电流的预期尖峰时间与实际尖峰时间之间的误差优化问题。SpikeProp算法在计算误差之前，将bienenstock - copper - munro (BCM)规则与Spike Timing Dependent Plasticity (STDP)相结合。隐层采用了突触权重关联训练(SWAT)中基于值的过滤思想。因此，在输入层和隐藏层之间的突触中使用时间选择器，这是通过兴奋性突触和抑制性突触的并行组合来实现的。在这两个过程中使用的神经元模型分别是Leaky Integrate and Fired (LIF)和Spike Response Model (SRM)。该算法针对非线性异或问题进行了基准测试。仿真结果显示了隐层时间选择器的框图和输出层测量的误差。
{"title":"A Parallel Combination of Facilitating Synapse Based on Temporal Correlation in SpikeProp Algorithm","authors":"Shushi Liu, Chuandong Li","doi":"10.1109/ICCSS53909.2021.9721950","DOIUrl":"https://doi.org/10.1109/ICCSS53909.2021.9721950","url":null,"abstract":"This paper presents a new method to minimize the error function between the expected spike time and the actual spike time, which is a parallel combination of facilitating synapses consisting of an excitatory and an inhibitory synapse. The SpikeProp algorithm is designed to solve the error optimization problem between the expected spike time and the actual spike time of the current from the presynaptic neuron passing through the synapse to the postsynaptic neuron. The SpikeProp algorithm merges the Bienenstock–Cooper–Munro (BCM) rule with Spike Timing Dependent Plasticity (STDP) before calculating errors. The idea of filtration based on value in Synaptic Weight Association Training (SWAT) is utilized in the hidden layer. Thus, a time selector is used in the synapse between the input layer and the hidden layer, which is achieved through parallel combination of excitatory and inhibitory synapses. The neuron models used in these two processes are Leaky Integrate and Fired (LIF) and Spike Response Model (SRM), respectively. The algorithm is benchmarked against the nonlinear exclusive OR (XOR) problem. The simulation results has illustrated the diagram of the time selector in the hidden layer and the error measured in the output layer.","PeriodicalId":435816,"journal":{"name":"2021 8th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS)","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129865584","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In the field of information security, it is very important to judge whether the information on a magnetic storage medium is completely destroyed. But so far, domestic research on the degaussing effect of magnetic storage media is still lacking. Previous studies have shown that the magnetic images before and after degaussing can reflect the amount of meaningful information left on the disk, which is closely related to the degaussing effect. Therefore, this paper proposes a new method to study the magnetic images before and after degaussing. This paper introduces the LBP texture feature extraction algorithm to process the magnetic images before and after degaussing, and evaluates the degaussing effect of the magnetic storage medium through the extracted texture feature values. A new LBP degaussing evaluation index is proposed, and the parameters of the index are optimized to achieve the best evaluation performance.
{"title":"LBP index for evaluation of disk degaussing achievement based on AFM image","authors":"Ziying Zhang, Zhe Xu, Yaxuan Yao, Xiaoge Liu, Jian Tang","doi":"10.1109/ICCSS53909.2021.9721960","DOIUrl":"https://doi.org/10.1109/ICCSS53909.2021.9721960","url":null,"abstract":"In the field of information security, it is very important to judge whether the information on a magnetic storage medium is completely destroyed. But so far, domestic research on the degaussing effect of magnetic storage media is still lacking. Previous studies have shown that the magnetic images before and after degaussing can reflect the amount of meaningful information left on the disk, which is closely related to the degaussing effect. Therefore, this paper proposes a new method to study the magnetic images before and after degaussing. This paper introduces the LBP texture feature extraction algorithm to process the magnetic images before and after degaussing, and evaluates the degaussing effect of the magnetic storage medium through the extracted texture feature values. A new LBP degaussing evaluation index is proposed, and the parameters of the index are optimized to achieve the best evaluation performance.","PeriodicalId":435816,"journal":{"name":"2021 8th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS)","volume":"179 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123004316","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-10DOI: 10.1109/ICCSS53909.2021.9721980
Feihong Xu, X. Luan
In the process of preparing xylenol, a large amount of tail gas will be generated. When using industrial boilers to treat the xylenol tail gas, the high-pressure steam in the furnace may lead to an explosion accident; on the other hand, the toxic tail gas of incomplete combustion in the furnace may also leak, endangering the lives of staff. So it is necessary to monitor the fault of the industrial boiler which used to treat xylenol tail gas. However, due to the non-stationary characteristics of the tail gas treatment process, the conventional fault monitoring methods have the problem of low accuracy. In order to solve these problems, this paper proposes a fault monitoring method based on trend similarity feature. This method cuts the time series by sliding time window, and calculates the trend similarity between data in each time window. Then uses the sliding time window to update the monitoring model in real-time. So it can change the threshold value of the monitoring model with the change of samples, to improve the monitoring accuracy. Finally, the practical data collected from a xylenol producer are used for validation. The results show that the fault detection based on the trend similarity feature has higher accuracy than the conventional method, and the detection accuracy increases with the non-stationary of the process.
{"title":"Trend similarity MWPCA based fault monitoring for xylenol tail gas treatment process","authors":"Feihong Xu, X. Luan","doi":"10.1109/ICCSS53909.2021.9721980","DOIUrl":"https://doi.org/10.1109/ICCSS53909.2021.9721980","url":null,"abstract":"In the process of preparing xylenol, a large amount of tail gas will be generated. When using industrial boilers to treat the xylenol tail gas, the high-pressure steam in the furnace may lead to an explosion accident; on the other hand, the toxic tail gas of incomplete combustion in the furnace may also leak, endangering the lives of staff. So it is necessary to monitor the fault of the industrial boiler which used to treat xylenol tail gas. However, due to the non-stationary characteristics of the tail gas treatment process, the conventional fault monitoring methods have the problem of low accuracy. In order to solve these problems, this paper proposes a fault monitoring method based on trend similarity feature. This method cuts the time series by sliding time window, and calculates the trend similarity between data in each time window. Then uses the sliding time window to update the monitoring model in real-time. So it can change the threshold value of the monitoring model with the change of samples, to improve the monitoring accuracy. Finally, the practical data collected from a xylenol producer are used for validation. The results show that the fault detection based on the trend similarity feature has higher accuracy than the conventional method, and the detection accuracy increases with the non-stationary of the process.","PeriodicalId":435816,"journal":{"name":"2021 8th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123863959","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-10DOI: 10.1109/ICCSS53909.2021.9721987
Yun Dai, Qing Yu, Tao Yang, Yuan Yao, Yi Liu
Development of reliable soft sensors using limited labeled samples is not an easy task in industrial processes. A selective generative adversarial network (SGAN)-based support vector regression (SGAN-SVR) soft sensor is proposed for quality prediction using limited labeled training data. Specifically, SVR is considered as a base prediction model. The Wasserstein GAN (WGAN) is adopted to capture the distribution of available labeled data and generate virtual candidates. Subsequently, using a proposed similarity measurement strategy, those synthetic data with more information are selected and introduced into the training set. Using the designed data augmentation approach, the SGAN-SVR model can achieve better prediction performance compared with the SVR soft sensor. The quality prediction results on an industrial polyethylene process demonstrate the effectiveness and advantages of the proposed method.
在工业过程中，使用有限的标记样品开发可靠的软传感器并不是一件容易的事。提出了一种基于选择性生成对抗网络(SGAN)的支持向量回归(SGAN- svr)软传感器，用于有限标记训练数据的质量预测。具体来说，SVR被认为是一个基本的预测模型。采用Wasserstein GAN (WGAN)捕获可用标记数据的分布并生成虚拟候选数据。然后，使用提出的相似度度量策略，选择具有更多信息的合成数据并将其引入训练集。采用设计的数据增强方法，与SVR软传感器相比，SGAN-SVR模型具有更好的预测性能。对某工业聚乙烯生产过程的质量预测结果表明了该方法的有效性和优越性。
{"title":"Enhanced Soft Sensor with Qualified Augmented Data Using Centroid Measurement Criterion","authors":"Yun Dai, Qing Yu, Tao Yang, Yuan Yao, Yi Liu","doi":"10.1109/ICCSS53909.2021.9721987","DOIUrl":"https://doi.org/10.1109/ICCSS53909.2021.9721987","url":null,"abstract":"Development of reliable soft sensors using limited labeled samples is not an easy task in industrial processes. A selective generative adversarial network (SGAN)-based support vector regression (SGAN-SVR) soft sensor is proposed for quality prediction using limited labeled training data. Specifically, SVR is considered as a base prediction model. The Wasserstein GAN (WGAN) is adopted to capture the distribution of available labeled data and generate virtual candidates. Subsequently, using a proposed similarity measurement strategy, those synthetic data with more information are selected and introduced into the training set. Using the designed data augmentation approach, the SGAN-SVR model can achieve better prediction performance compared with the SVR soft sensor. The quality prediction results on an industrial polyethylene process demonstrate the effectiveness and advantages of the proposed method.","PeriodicalId":435816,"journal":{"name":"2021 8th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128128030","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-10DOI: 10.1109/ICCSS53909.2021.9722011
Ruirong Lin, Nangfeng Xiao
Compared with single image super-resolution (SISR), reference-based image super-resolution (RefSR) utilizes additional references (Ref) to recover more realistic texture details, achieving better reconstruction performance. Most recent works focus on transferring relevant texture features from Ref to low-resolution (LR) images. However, those works ignore the high-frequency information existing in the LR space, leading to performance degradation when irrelevant Ref images are given. To address this issue, we propose a residual channel attention connection network for reference-based image super-resolution (RCACSR), which fuses valuable high-frequency information in LR space with high-resolution (HR) texture details of Ref. Specifically, the proposed residual channel attention connection network (RCACN) can extract more complex features from the LR space. Moreover, an enhanced texture transformer is presented, which can search and transfer texture features more accurately from Ref. Extensive experiments have demonstrated that the proposed RCACSR is superior to the state-of-the-art approaches in the aspects of both quantitative and qualitative measurements.
{"title":"Residual Channel Attention Connection Network for Reference-based Image Super-resolution","authors":"Ruirong Lin, Nangfeng Xiao","doi":"10.1109/ICCSS53909.2021.9722011","DOIUrl":"https://doi.org/10.1109/ICCSS53909.2021.9722011","url":null,"abstract":"Compared with single image super-resolution (SISR), reference-based image super-resolution (RefSR) utilizes additional references (Ref) to recover more realistic texture details, achieving better reconstruction performance. Most recent works focus on transferring relevant texture features from Ref to low-resolution (LR) images. However, those works ignore the high-frequency information existing in the LR space, leading to performance degradation when irrelevant Ref images are given. To address this issue, we propose a residual channel attention connection network for reference-based image super-resolution (RCACSR), which fuses valuable high-frequency information in LR space with high-resolution (HR) texture details of Ref. Specifically, the proposed residual channel attention connection network (RCACN) can extract more complex features from the LR space. Moreover, an enhanced texture transformer is presented, which can search and transfer texture features more accurately from Ref. Extensive experiments have demonstrated that the proposed RCACSR is superior to the state-of-the-art approaches in the aspects of both quantitative and qualitative measurements.","PeriodicalId":435816,"journal":{"name":"2021 8th International Conference on Information, Cybernetics, and Computational Social Systems (ICCSS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131265570","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：