This paper presents an off-line handwritten signature verification system based on the Siamese network,where a hybrid architecture is used.The Residual neural Network(ResNet)is used to realize a powerful feature extra...This paper presents an off-line handwritten signature verification system based on the Siamese network,where a hybrid architecture is used.The Residual neural Network(ResNet)is used to realize a powerful feature extraction model such that Writer Independent(WI)features can be effectively learned.A single-layer Siamese Neural Network(NN)is used to realize a Writer Dependent(WD)classifier such that the storage space can be minimized.For the purpose of reducing the impact of the high intraclass variability of the signature and ensuring that the Siamese network can learn more effectively,we propose a method of selecting a reference signature as one of the inputs for the Siamese network.To take full advantage of the reference signature,we modify the conventional contrastive loss function to enhance the accuracy.By using the proposed techniques,the accuracy of the system can be increased by 5.9%.Based on the GPDS signature dataset,the proposed system is able to achieve an accuracy of 94.61%which is better than the accuracy achieved by the current state-of-the-art work.展开更多
This article addresses the problem of reference picture optimization in video communication over error prone networks. A novel estimation model for transmission distortion is proposed. This model is capable of recursi...This article addresses the problem of reference picture optimization in video communication over error prone networks. A novel estimation model for transmission distortion is proposed. This model is capable of recursively estimating the overall end-to-end distortion caused by quantization, error propagation, and error concealment. Simulation results show that this model can accurately estimate channel distortion. Then, based on the distortion estimation model, a new non-feedback key-frame reference picture selection （KRPS） algorithm is developed. The optimum reference picture minimizes the transmission distortion under the rate-distortion optimization framework. Extensive experiment results demonstrate that the proposed KRPS algorithm substantially achieves more peak signal to noise ratio （PSNR） gain over traditional prediction, especially in low bit-rate transmission.展开更多
Quality degradation occurs during transmission of video streaming over the error-prone network. By jointly using redundant slice, reference frame selection, and intra/inters mode decision, a content and end-to-end rat...Quality degradation occurs during transmission of video streaming over the error-prone network. By jointly using redundant slice, reference frame selection, and intra/inters mode decision, a content and end-to-end rate-distortion based error resilience method is proposed. Firstly, the intra/inter mode decision is implemented using macro-block(MB) refresh, and then redundant picture and reference frame selection are utilized together to realize the redundant coding. The estimated error propagation distortion and bit consumption of refresh MB are used for the mode and reference frame decision of refresh MB. Secondly, by analyzing the statistical property in the successive frames, the error propagation distortion and bit consumption are formulated as a function of temporal distance. Encoding parameters of the current frame is determined by the estimated error propagation distortion and bit consumption. Thirdly, by comparing the rate-distortion cost of different combinations, proper selection of error resilience method is performed before the encoding process of the current frame. Finally, the MB mode and bit distribution of the primary picture are analyzed for the derivation of the texture information. The motion information is subsequently incorporated for the calculation of video content complexity to implement the content based redundant coding. Experimental results demonstrate that the proposed algorithm achieves significant performance gains over the LA-RDO and HRP method when video is transmitted over error-prone channel.展开更多
lsochrysis zhangfiangensis is a potential marine microalga for biodiesel production, which accumulates lipid under ni- trogen limitation conditions, but the mechanism on molecular level is veiled. Quantitative real-ti...lsochrysis zhangfiangensis is a potential marine microalga for biodiesel production, which accumulates lipid under ni- trogen limitation conditions, but the mechanism on molecular level is veiled. Quantitative real-time polymerase chain reaction （qPCR） provides the possibility to investigate the gene expression levels, and a valid reference for data normalization is an essential prerequisite for firing up the analysis. In this study, five housekeeping genes, actin （ACT）, α-tubulin （TUA）, β-tubulin （TUB）, ubiquitin （UBI）, 18S rRNA （18S） and one target gene, diacylglycerol acyltransferase （DGAT）, were used for determining the reference. By analyzing the stabilities based on calculation of the stability index and on operating the two types of software, geNorm and bestkeeper, it showed that the reference genes widely used in higher plant and microalgae, such as UBI, TUA and 18S, were not the most stable ones in nitrogen-stressed 1. zhangjiangensis, and thus are not suitable for exploring the mRNA expression levels under these experi- mental conditions. Our results show that ACT together with TUB is the most feasible internal control for investigating gene expres- sion under nitrogen-stressed conditions. Our findings will contribute not only to future qPCR studies of/. zhangfiangensis, but also to verification of comparative transcriptomics studies of the microalgae under similar conditions.展开更多
Obtaining the location of an unknown node accurately is a key problem of a locating service under a ubiquitous computing environment.The paper proposes and proves three theorems of location reference node placement ac...Obtaining the location of an unknown node accurately is a key problem of a locating service under a ubiquitous computing environment.The paper proposes and proves three theorems of location reference node placement according to the analysis of the location error produced during location using a polygon location method and three important characteristics of chaos dynamics.Based on the three theorems,the location reference node selection(LRNS)algorithm is proposed by improving on the traditional polygon location algorithm.The simulation results indicate that the reference node placement theorems and the LRNS algo-rithm can meet the requirements of a ubiquitous terminal’s real-time location and possess a preferable precision in location.展开更多
This paper presents an automatic techruque of suitable reference potential selection for quantitative EEG interpretation.The 16-channels EEG recording under mono-polar derivation is analyzed.There are two prejudgments...This paper presents an automatic techruque of suitable reference potential selection for quantitative EEG interpretation.The 16-channels EEG recording under mono-polar derivation is analyzed.There are two prejudgments defined for checking the amplitude distribution and ear lobe activation.After prejudgments,the EEG is classified into several cases in cluding diffused case,non-diffused case,and artifact contami nation case.Due to the cases,an automatic reference selection method is applied in order to find out suitable reference potential.Finally,the referential derivation constructed according to the obtained reference potential,is evaluated for further EEG rhythm analysis.The presented technique can high light the EEG rhythm of interest,which is useful for quantitative EEG interpretation by both visual inspection and automatic evaluation.展开更多
文摘This paper presents an off-line handwritten signature verification system based on the Siamese network,where a hybrid architecture is used.The Residual neural Network(ResNet)is used to realize a powerful feature extraction model such that Writer Independent(WI)features can be effectively learned.A single-layer Siamese Neural Network(NN)is used to realize a Writer Dependent(WD)classifier such that the storage space can be minimized.For the purpose of reducing the impact of the high intraclass variability of the signature and ensuring that the Siamese network can learn more effectively,we propose a method of selecting a reference signature as one of the inputs for the Siamese network.To take full advantage of the reference signature,we modify the conventional contrastive loss function to enhance the accuracy.By using the proposed techniques,the accuracy of the system can be increased by 5.9%.Based on the GPDS signature dataset,the proposed system is able to achieve an accuracy of 94.61%which is better than the accuracy achieved by the current state-of-the-art work.
基金supported by the National Natural Science Foundation of China (60672099)
文摘This article addresses the problem of reference picture optimization in video communication over error prone networks. A novel estimation model for transmission distortion is proposed. This model is capable of recursively estimating the overall end-to-end distortion caused by quantization, error propagation, and error concealment. Simulation results show that this model can accurately estimate channel distortion. Then, based on the distortion estimation model, a new non-feedback key-frame reference picture selection （KRPS） algorithm is developed. The optimum reference picture minimizes the transmission distortion under the rate-distortion optimization framework. Extensive experiment results demonstrate that the proposed KRPS algorithm substantially achieves more peak signal to noise ratio （PSNR） gain over traditional prediction, especially in low bit-rate transmission.
基金Project(40927001)supported by the National Natural Science Foundation of ChinaProject(2011R09021-06)supported by the Program of Key Scientific and Technological Innovation Team of Zhejiang Province,ChinaProject supported by the Fundamental Research Funds for the Central Universities of China
文摘Quality degradation occurs during transmission of video streaming over the error-prone network. By jointly using redundant slice, reference frame selection, and intra/inters mode decision, a content and end-to-end rate-distortion based error resilience method is proposed. Firstly, the intra/inter mode decision is implemented using macro-block(MB) refresh, and then redundant picture and reference frame selection are utilized together to realize the redundant coding. The estimated error propagation distortion and bit consumption of refresh MB are used for the mode and reference frame decision of refresh MB. Secondly, by analyzing the statistical property in the successive frames, the error propagation distortion and bit consumption are formulated as a function of temporal distance. Encoding parameters of the current frame is determined by the estimated error propagation distortion and bit consumption. Thirdly, by comparing the rate-distortion cost of different combinations, proper selection of error resilience method is performed before the encoding process of the current frame. Finally, the MB mode and bit distribution of the primary picture are analyzed for the derivation of the texture information. The motion information is subsequently incorporated for the calculation of video content complexity to implement the content based redundant coding. Experimental results demonstrate that the proposed algorithm achieves significant performance gains over the LA-RDO and HRP method when video is transmitted over error-prone channel.
基金supported by the Hundred Talent Program of the Chinese Academy of Sciences (No.A1097)National High Technology Research and Development Program ‘863’ (2012AA052101)+2 种基金Liaoning Provincial Natural Science Foundation (2012010263)Knowledge Innovation Programs of Dalian Institute of Chemical Physics,CAS (K2010A13)China Postdoctoral Science Foundation Funded Project (2014M551139)
文摘lsochrysis zhangfiangensis is a potential marine microalga for biodiesel production, which accumulates lipid under ni- trogen limitation conditions, but the mechanism on molecular level is veiled. Quantitative real-time polymerase chain reaction （qPCR） provides the possibility to investigate the gene expression levels, and a valid reference for data normalization is an essential prerequisite for firing up the analysis. In this study, five housekeeping genes, actin （ACT）, α-tubulin （TUA）, β-tubulin （TUB）, ubiquitin （UBI）, 18S rRNA （18S） and one target gene, diacylglycerol acyltransferase （DGAT）, were used for determining the reference. By analyzing the stabilities based on calculation of the stability index and on operating the two types of software, geNorm and bestkeeper, it showed that the reference genes widely used in higher plant and microalgae, such as UBI, TUA and 18S, were not the most stable ones in nitrogen-stressed 1. zhangjiangensis, and thus are not suitable for exploring the mRNA expression levels under these experi- mental conditions. Our results show that ACT together with TUB is the most feasible internal control for investigating gene expres- sion under nitrogen-stressed conditions. Our findings will contribute not only to future qPCR studies of/. zhangfiangensis, but also to verification of comparative transcriptomics studies of the microalgae under similar conditions.
基金supported by the National Natural Science Foundation of China(Grant No.69873007)the Hi-Tech Research and Development Program of China(No.2001AA415320).
文摘Obtaining the location of an unknown node accurately is a key problem of a locating service under a ubiquitous computing environment.The paper proposes and proves three theorems of location reference node placement according to the analysis of the location error produced during location using a polygon location method and three important characteristics of chaos dynamics.Based on the three theorems,the location reference node selection(LRNS)algorithm is proposed by improving on the traditional polygon location algorithm.The simulation results indicate that the reference node placement theorems and the LRNS algo-rithm can meet the requirements of a ubiquitous terminal’s real-time location and possess a preferable precision in location.
基金Grant sponsor:National Natural Science Foundation of China,grant number:61074113grant sponsor:Shanghai Leading Academic Discipline Project,grant number:B504grant sponsor:Fundamental Research Funds for the Central Universities,grant number:WH0914028
文摘This paper presents an automatic techruque of suitable reference potential selection for quantitative EEG interpretation.The 16-channels EEG recording under mono-polar derivation is analyzed.There are two prejudgments defined for checking the amplitude distribution and ear lobe activation.After prejudgments,the EEG is classified into several cases in cluding diffused case,non-diffused case,and artifact contami nation case.Due to the cases,an automatic reference selection method is applied in order to find out suitable reference potential.Finally,the referential derivation constructed according to the obtained reference potential,is evaluated for further EEG rhythm analysis.The presented technique can high light the EEG rhythm of interest,which is useful for quantitative EEG interpretation by both visual inspection and automatic evaluation.
